Moved to Visual Studio 2015, OpenCV 3.2.0, x64

include/opencv2/core/affine.hpp

@@ -0,0 +1,517 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                          License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef OPENCV_CORE_AFFINE3_HPP
+#define OPENCV_CORE_AFFINE3_HPP
+
+#ifdef __cplusplus
+
+#include <opencv2/core.hpp>
+
+namespace cv
+{
+
+//! @addtogroup core
+//! @{
+
+    /** @brief Affine transform
+      @todo document
+     */
+    template<typename T>
+    class Affine3
+    {
+    public:
+        typedef T float_type;
+        typedef Matx<float_type, 3, 3> Mat3;
+        typedef Matx<float_type, 4, 4> Mat4;
+        typedef Vec<float_type, 3> Vec3;
+
+        Affine3();
+
+        //! Augmented affine matrix
+        Affine3(const Mat4& affine);
+
+        //! Rotation matrix
+        Affine3(const Mat3& R, const Vec3& t = Vec3::all(0));
+
+        //! Rodrigues vector
+        Affine3(const Vec3& rvec, const Vec3& t = Vec3::all(0));
+
+        //! Combines all contructors above. Supports 4x4, 4x3, 3x3, 1x3, 3x1 sizes of data matrix
+        explicit Affine3(const Mat& data, const Vec3& t = Vec3::all(0));
+
+        //! From 16th element array
+        explicit Affine3(const float_type* vals);
+
+        //! Create identity transform
+        static Affine3 Identity();
+
+        //! Rotation matrix
+        void rotation(const Mat3& R);
+
+        //! Rodrigues vector
+        void rotation(const Vec3& rvec);
+
+        //! Combines rotation methods above. Suports 3x3, 1x3, 3x1 sizes of data matrix;
+        void rotation(const Mat& data);
+
+        void linear(const Mat3& L);
+        void translation(const Vec3& t);
+
+        Mat3 rotation() const;
+        Mat3 linear() const;
+        Vec3 translation() const;
+
+        //! Rodrigues vector
+        Vec3 rvec() const;
+
+        Affine3 inv(int method = cv::DECOMP_SVD) const;
+
+        //! a.rotate(R) is equivalent to Affine(R, 0) * a;
+        Affine3 rotate(const Mat3& R) const;
+
+        //! a.rotate(rvec) is equivalent to Affine(rvec, 0) * a;
+        Affine3 rotate(const Vec3& rvec) const;
+
+        //! a.translate(t) is equivalent to Affine(E, t) * a;
+        Affine3 translate(const Vec3& t) const;
+
+        //! a.concatenate(affine) is equivalent to affine * a;
+        Affine3 concatenate(const Affine3& affine) const;
+
+        template <typename Y> operator Affine3<Y>() const;
+
+        template <typename Y> Affine3<Y> cast() const;
+
+        Mat4 matrix;
+
+#if defined EIGEN_WORLD_VERSION && defined EIGEN_GEOMETRY_MODULE_H
+        Affine3(const Eigen::Transform<T, 3, Eigen::Affine, (Eigen::RowMajor)>& affine);
+        Affine3(const Eigen::Transform<T, 3, Eigen::Affine>& affine);
+        operator Eigen::Transform<T, 3, Eigen::Affine, (Eigen::RowMajor)>() const;
+        operator Eigen::Transform<T, 3, Eigen::Affine>() const;
+#endif
+    };
+
+    template<typename T> static
+    Affine3<T> operator*(const Affine3<T>& affine1, const Affine3<T>& affine2);
+
+    template<typename T, typename V> static
+    V operator*(const Affine3<T>& affine, const V& vector);
+
+    typedef Affine3<float> Affine3f;
+    typedef Affine3<double> Affine3d;
+
+    static Vec3f operator*(const Affine3f& affine, const Vec3f& vector);
+    static Vec3d operator*(const Affine3d& affine, const Vec3d& vector);
+
+    template<typename _Tp> class DataType< Affine3<_Tp> >
+    {
+    public:
+        typedef Affine3<_Tp>                               value_type;
+        typedef Affine3<typename DataType<_Tp>::work_type> work_type;
+        typedef _Tp                                        channel_type;
+
+        enum { generic_type = 0,
+               depth        = DataType<channel_type>::depth,
+               channels     = 16,
+               fmt          = DataType<channel_type>::fmt + ((channels - 1) << 8),
+               type         = CV_MAKETYPE(depth, channels)
+             };
+
+        typedef Vec<channel_type, channels> vec_type;
+    };
+
+//! @} core
+
+}
+
+//! @cond IGNORED
+
+///////////////////////////////////////////////////////////////////////////////////
+// Implementaiton
+
+template<typename T> inline
+cv::Affine3<T>::Affine3()
+    : matrix(Mat4::eye())
+{}
+
+template<typename T> inline
+cv::Affine3<T>::Affine3(const Mat4& affine)
+    : matrix(affine)
+{}
+
+template<typename T> inline
+cv::Affine3<T>::Affine3(const Mat3& R, const Vec3& t)
+{
+    rotation(R);
+    translation(t);
+    matrix.val[12] = matrix.val[13] = matrix.val[14] = 0;
+    matrix.val[15] = 1;
+}
+
+template<typename T> inline
+cv::Affine3<T>::Affine3(const Vec3& _rvec, const Vec3& t)
+{
+    rotation(_rvec);
+    translation(t);
+    matrix.val[12] = matrix.val[13] = matrix.val[14] = 0;
+    matrix.val[15] = 1;
+}
+
+template<typename T> inline
+cv::Affine3<T>::Affine3(const cv::Mat& data, const Vec3& t)
+{
+    CV_Assert(data.type() == cv::DataType<T>::type);
+
+    if (data.cols == 4 && data.rows == 4)
+    {
+        data.copyTo(matrix);
+        return;
+    }
+    else if (data.cols == 4 && data.rows == 3)
+    {
+        rotation(data(Rect(0, 0, 3, 3)));
+        translation(data(Rect(3, 0, 1, 3)));
+        return;
+    }
+
+    rotation(data);
+    translation(t);
+    matrix.val[12] = matrix.val[13] = matrix.val[14] = 0;
+    matrix.val[15] = 1;
+}
+
+template<typename T> inline
+cv::Affine3<T>::Affine3(const float_type* vals) : matrix(vals)
+{}
+
+template<typename T> inline
+cv::Affine3<T> cv::Affine3<T>::Identity()
+{
+    return Affine3<T>(cv::Affine3<T>::Mat4::eye());
+}
+
+template<typename T> inline
+void cv::Affine3<T>::rotation(const Mat3& R)
+{
+    linear(R);
+}
+
+template<typename T> inline
+void cv::Affine3<T>::rotation(const Vec3& _rvec)
+{
+    double theta = norm(_rvec);
+
+    if (theta < DBL_EPSILON)
+        rotation(Mat3::eye());
+    else
+    {
+        double c = std::cos(theta);
+        double s = std::sin(theta);
+        double c1 = 1. - c;
+        double itheta = (theta != 0) ? 1./theta : 0.;
+
+        Point3_<T> r = _rvec*itheta;
+
+        Mat3 rrt( r.x*r.x, r.x*r.y, r.x*r.z, r.x*r.y, r.y*r.y, r.y*r.z, r.x*r.z, r.y*r.z, r.z*r.z );
+        Mat3 r_x( 0, -r.z, r.y, r.z, 0, -r.x, -r.y, r.x, 0 );
+
+        // R = cos(theta)*I + (1 - cos(theta))*r*rT + sin(theta)*[r_x]
+        // where [r_x] is [0 -rz ry; rz 0 -rx; -ry rx 0]
+        Mat3 R = c*Mat3::eye() + c1*rrt + s*r_x;
+
+        rotation(R);
+    }
+}
+
+//Combines rotation methods above. Suports 3x3, 1x3, 3x1 sizes of data matrix;
+template<typename T> inline
+void cv::Affine3<T>::rotation(const cv::Mat& data)
+{
+    CV_Assert(data.type() == cv::DataType<T>::type);
+
+    if (data.cols == 3 && data.rows == 3)
+    {
+        Mat3 R;
+        data.copyTo(R);
+        rotation(R);
+    }
+    else if ((data.cols == 3 && data.rows == 1) || (data.cols == 1 && data.rows == 3))
+    {
+        Vec3 _rvec;
+        data.reshape(1, 3).copyTo(_rvec);
+        rotation(_rvec);
+    }
+    else
+        CV_Assert(!"Input marix can be 3x3, 1x3 or 3x1");
+}
+
+template<typename T> inline
+void cv::Affine3<T>::linear(const Mat3& L)
+{
+    matrix.val[0] = L.val[0]; matrix.val[1] = L.val[1];  matrix.val[ 2] = L.val[2];
+    matrix.val[4] = L.val[3]; matrix.val[5] = L.val[4];  matrix.val[ 6] = L.val[5];
+    matrix.val[8] = L.val[6]; matrix.val[9] = L.val[7];  matrix.val[10] = L.val[8];
+}
+
+template<typename T> inline
+void cv::Affine3<T>::translation(const Vec3& t)
+{
+    matrix.val[3] = t[0]; matrix.val[7] = t[1]; matrix.val[11] = t[2];
+}
+
+template<typename T> inline
+typename cv::Affine3<T>::Mat3 cv::Affine3<T>::rotation() const
+{
+    return linear();
+}
+
+template<typename T> inline
+typename cv::Affine3<T>::Mat3 cv::Affine3<T>::linear() const
+{
+    typename cv::Affine3<T>::Mat3 R;
+    R.val[0] = matrix.val[0];  R.val[1] = matrix.val[1];  R.val[2] = matrix.val[ 2];
+    R.val[3] = matrix.val[4];  R.val[4] = matrix.val[5];  R.val[5] = matrix.val[ 6];
+    R.val[6] = matrix.val[8];  R.val[7] = matrix.val[9];  R.val[8] = matrix.val[10];
+    return R;
+}
+
+template<typename T> inline
+typename cv::Affine3<T>::Vec3 cv::Affine3<T>::translation() const
+{
+    return Vec3(matrix.val[3], matrix.val[7], matrix.val[11]);
+}
+
+template<typename T> inline
+typename cv::Affine3<T>::Vec3 cv::Affine3<T>::rvec() const
+{
+    cv::Vec3d w;
+    cv::Matx33d u, vt, R = rotation();
+    cv::SVD::compute(R, w, u, vt, cv::SVD::FULL_UV + cv::SVD::MODIFY_A);
+    R = u * vt;
+
+    double rx = R.val[7] - R.val[5];
+    double ry = R.val[2] - R.val[6];
+    double rz = R.val[3] - R.val[1];
+
+    double s = std::sqrt((rx*rx + ry*ry + rz*rz)*0.25);
+    double c = (R.val[0] + R.val[4] + R.val[8] - 1) * 0.5;
+    c = c > 1.0 ? 1.0 : c < -1.0 ? -1.0 : c;
+    double theta = acos(c);
+
+    if( s < 1e-5 )
+    {
+        if( c > 0 )
+            rx = ry = rz = 0;
+        else
+        {
+            double t;
+            t = (R.val[0] + 1) * 0.5;
+            rx = std::sqrt(std::max(t, 0.0));
+            t = (R.val[4] + 1) * 0.5;
+            ry = std::sqrt(std::max(t, 0.0)) * (R.val[1] < 0 ? -1.0 : 1.0);
+            t = (R.val[8] + 1) * 0.5;
+            rz = std::sqrt(std::max(t, 0.0)) * (R.val[2] < 0 ? -1.0 : 1.0);
+
+            if( fabs(rx) < fabs(ry) && fabs(rx) < fabs(rz) && (R.val[5] > 0) != (ry*rz > 0) )
+                rz = -rz;
+            theta /= std::sqrt(rx*rx + ry*ry + rz*rz);
+            rx *= theta;
+            ry *= theta;
+            rz *= theta;
+        }
+    }
+    else
+    {
+        double vth = 1/(2*s);
+        vth *= theta;
+        rx *= vth; ry *= vth; rz *= vth;
+    }
+
+    return cv::Vec3d(rx, ry, rz);
+}
+
+template<typename T> inline
+cv::Affine3<T> cv::Affine3<T>::inv(int method) const
+{
+    return matrix.inv(method);
+}
+
+template<typename T> inline
+cv::Affine3<T> cv::Affine3<T>::rotate(const Mat3& R) const
+{
+    Mat3 Lc = linear();
+    Vec3 tc = translation();
+    Mat4 result;
+    result.val[12] = result.val[13] = result.val[14] = 0;
+    result.val[15] = 1;
+
+    for(int j = 0; j < 3; ++j)
+    {
+        for(int i = 0; i < 3; ++i)
+        {
+            float_type value = 0;
+            for(int k = 0; k < 3; ++k)
+                value += R(j, k) * Lc(k, i);
+            result(j, i) = value;
+        }
+
+        result(j, 3) = R.row(j).dot(tc.t());
+    }
+    return result;
+}
+
+template<typename T> inline
+cv::Affine3<T> cv::Affine3<T>::rotate(const Vec3& _rvec) const
+{
+    return rotate(Affine3f(_rvec).rotation());
+}
+
+template<typename T> inline
+cv::Affine3<T> cv::Affine3<T>::translate(const Vec3& t) const
+{
+    Mat4 m = matrix;
+    m.val[ 3] += t[0];
+    m.val[ 7] += t[1];
+    m.val[11] += t[2];
+    return m;
+}
+
+template<typename T> inline
+cv::Affine3<T> cv::Affine3<T>::concatenate(const Affine3<T>& affine) const
+{
+    return (*this).rotate(affine.rotation()).translate(affine.translation());
+}
+
+template<typename T> template <typename Y> inline
+cv::Affine3<T>::operator Affine3<Y>() const
+{
+    return Affine3<Y>(matrix);
+}
+
+template<typename T> template <typename Y> inline
+cv::Affine3<Y> cv::Affine3<T>::cast() const
+{
+    return Affine3<Y>(matrix);
+}
+
+template<typename T> inline
+cv::Affine3<T> cv::operator*(const cv::Affine3<T>& affine1, const cv::Affine3<T>& affine2)
+{
+    return affine2.concatenate(affine1);
+}
+
+template<typename T, typename V> inline
+V cv::operator*(const cv::Affine3<T>& affine, const V& v)
+{
+    const typename Affine3<T>::Mat4& m = affine.matrix;
+
+    V r;
+    r.x = m.val[0] * v.x + m.val[1] * v.y + m.val[ 2] * v.z + m.val[ 3];
+    r.y = m.val[4] * v.x + m.val[5] * v.y + m.val[ 6] * v.z + m.val[ 7];
+    r.z = m.val[8] * v.x + m.val[9] * v.y + m.val[10] * v.z + m.val[11];
+    return r;
+}
+
+static inline
+cv::Vec3f cv::operator*(const cv::Affine3f& affine, const cv::Vec3f& v)
+{
+    const cv::Matx44f& m = affine.matrix;
+    cv::Vec3f r;
+    r.val[0] = m.val[0] * v[0] + m.val[1] * v[1] + m.val[ 2] * v[2] + m.val[ 3];
+    r.val[1] = m.val[4] * v[0] + m.val[5] * v[1] + m.val[ 6] * v[2] + m.val[ 7];
+    r.val[2] = m.val[8] * v[0] + m.val[9] * v[1] + m.val[10] * v[2] + m.val[11];
+    return r;
+}
+
+static inline
+cv::Vec3d cv::operator*(const cv::Affine3d& affine, const cv::Vec3d& v)
+{
+    const cv::Matx44d& m = affine.matrix;
+    cv::Vec3d r;
+    r.val[0] = m.val[0] * v[0] + m.val[1] * v[1] + m.val[ 2] * v[2] + m.val[ 3];
+    r.val[1] = m.val[4] * v[0] + m.val[5] * v[1] + m.val[ 6] * v[2] + m.val[ 7];
+    r.val[2] = m.val[8] * v[0] + m.val[9] * v[1] + m.val[10] * v[2] + m.val[11];
+    return r;
+}
+
+
+
+#if defined EIGEN_WORLD_VERSION && defined EIGEN_GEOMETRY_MODULE_H
+
+template<typename T> inline
+cv::Affine3<T>::Affine3(const Eigen::Transform<T, 3, Eigen::Affine, (Eigen::RowMajor)>& affine)
+{
+    cv::Mat(4, 4, cv::DataType<T>::type, affine.matrix().data()).copyTo(matrix);
+}
+
+template<typename T> inline
+cv::Affine3<T>::Affine3(const Eigen::Transform<T, 3, Eigen::Affine>& affine)
+{
+    Eigen::Transform<T, 3, Eigen::Affine, (Eigen::RowMajor)> a = affine;
+    cv::Mat(4, 4, cv::DataType<T>::type, a.matrix().data()).copyTo(matrix);
+}
+
+template<typename T> inline
+cv::Affine3<T>::operator Eigen::Transform<T, 3, Eigen::Affine, (Eigen::RowMajor)>() const
+{
+    Eigen::Transform<T, 3, Eigen::Affine, (Eigen::RowMajor)> r;
+    cv::Mat hdr(4, 4, cv::DataType<T>::type, r.matrix().data());
+    cv::Mat(matrix, false).copyTo(hdr);
+    return r;
+}
+
+template<typename T> inline
+cv::Affine3<T>::operator Eigen::Transform<T, 3, Eigen::Affine>() const
+{
+    return this->operator Eigen::Transform<T, 3, Eigen::Affine, (Eigen::RowMajor)>();
+}
+
+#endif /* defined EIGEN_WORLD_VERSION && defined EIGEN_GEOMETRY_MODULE_H */
+
+//! @endcond
+
+#endif /* __cplusplus */
+
+#endif /* OPENCV_CORE_AFFINE3_HPP */

include/opencv2/core/base.hpp

@@ -0,0 +1,691 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                          License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
+// Copyright (C) 2014, Itseez Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef OPENCV_CORE_BASE_HPP
+#define OPENCV_CORE_BASE_HPP
+
+#ifndef __cplusplus
+#  error base.hpp header must be compiled as C++
+#endif
+
+#include "opencv2/opencv_modules.hpp"
+
+#include <climits>
+#include <algorithm>
+
+#include "opencv2/core/cvdef.h"
+#include "opencv2/core/cvstd.hpp"
+
+namespace cv
+{
+
+//! @addtogroup core_utils
+//! @{
+
+namespace Error {
+//! error codes
+enum Code {
+    StsOk=                       0,  //!< everithing is ok
+    StsBackTrace=               -1,  //!< pseudo error for back trace
+    StsError=                   -2,  //!< unknown /unspecified error
+    StsInternal=                -3,  //!< internal error (bad state)
+    StsNoMem=                   -4,  //!< insufficient memory
+    StsBadArg=                  -5,  //!< function arg/param is bad
+    StsBadFunc=                 -6,  //!< unsupported function
+    StsNoConv=                  -7,  //!< iter. didn't converge
+    StsAutoTrace=               -8,  //!< tracing
+    HeaderIsNull=               -9,  //!< image header is NULL
+    BadImageSize=              -10,  //!< image size is invalid
+    BadOffset=                 -11,  //!< offset is invalid
+    BadDataPtr=                -12,  //!<
+    BadStep=                   -13,  //!<
+    BadModelOrChSeq=           -14,  //!<
+    BadNumChannels=            -15,  //!<
+    BadNumChannel1U=           -16,  //!<
+    BadDepth=                  -17,  //!<
+    BadAlphaChannel=           -18,  //!<
+    BadOrder=                  -19,  //!<
+    BadOrigin=                 -20,  //!<
+    BadAlign=                  -21,  //!<
+    BadCallBack=               -22,  //!<
+    BadTileSize=               -23,  //!<
+    BadCOI=                    -24,  //!<
+    BadROISize=                -25,  //!<
+    MaskIsTiled=               -26,  //!<
+    StsNullPtr=                -27,  //!< null pointer
+    StsVecLengthErr=           -28,  //!< incorrect vector length
+    StsFilterStructContentErr= -29,  //!< incorr. filter structure content
+    StsKernelStructContentErr= -30,  //!< incorr. transform kernel content
+    StsFilterOffsetErr=        -31,  //!< incorrect filter ofset value
+    StsBadSize=                -201, //!< the input/output structure size is incorrect
+    StsDivByZero=              -202, //!< division by zero
+    StsInplaceNotSupported=    -203, //!< in-place operation is not supported
+    StsObjectNotFound=         -204, //!< request can't be completed
+    StsUnmatchedFormats=       -205, //!< formats of input/output arrays differ
+    StsBadFlag=                -206, //!< flag is wrong or not supported
+    StsBadPoint=               -207, //!< bad CvPoint
+    StsBadMask=                -208, //!< bad format of mask (neither 8uC1 nor 8sC1)
+    StsUnmatchedSizes=         -209, //!< sizes of input/output structures do not match
+    StsUnsupportedFormat=      -210, //!< the data format/type is not supported by the function
+    StsOutOfRange=             -211, //!< some of parameters are out of range
+    StsParseError=             -212, //!< invalid syntax/structure of the parsed file
+    StsNotImplemented=         -213, //!< the requested function/feature is not implemented
+    StsBadMemBlock=            -214, //!< an allocated block has been corrupted
+    StsAssert=                 -215, //!< assertion failed
+    GpuNotSupported=           -216,
+    GpuApiCallError=           -217,
+    OpenGlNotSupported=        -218,
+    OpenGlApiCallError=        -219,
+    OpenCLApiCallError=        -220,
+    OpenCLDoubleNotSupported=  -221,
+    OpenCLInitError=           -222,
+    OpenCLNoAMDBlasFft=        -223
+};
+} //Error
+
+//! @} core_utils
+
+//! @addtogroup core_array
+//! @{
+
+//! matrix decomposition types
+enum DecompTypes {
+    /** Gaussian elimination with the optimal pivot element chosen. */
+    DECOMP_LU       = 0,
+    /** singular value decomposition (SVD) method; the system can be over-defined and/or the matrix
+    src1 can be singular */
+    DECOMP_SVD      = 1,
+    /** eigenvalue decomposition; the matrix src1 must be symmetrical */
+    DECOMP_EIG      = 2,
+    /** Cholesky \f$LL^T\f$ factorization; the matrix src1 must be symmetrical and positively
+    defined */
+    DECOMP_CHOLESKY = 3,
+    /** QR factorization; the system can be over-defined and/or the matrix src1 can be singular */
+    DECOMP_QR       = 4,
+    /** while all the previous flags are mutually exclusive, this flag can be used together with
+    any of the previous; it means that the normal equations
+    \f$\texttt{src1}^T\cdot\texttt{src1}\cdot\texttt{dst}=\texttt{src1}^T\texttt{src2}\f$ are
+    solved instead of the original system
+    \f$\texttt{src1}\cdot\texttt{dst}=\texttt{src2}\f$ */
+    DECOMP_NORMAL   = 16
+};
+
+/** norm types
+- For one array:
+\f[norm =  \forkthree{\|\texttt{src1}\|_{L_{\infty}} =  \max _I | \texttt{src1} (I)|}{if  \(\texttt{normType} = \texttt{NORM_INF}\) }
+{ \| \texttt{src1} \| _{L_1} =  \sum _I | \texttt{src1} (I)|}{if  \(\texttt{normType} = \texttt{NORM_L1}\) }
+{ \| \texttt{src1} \| _{L_2} =  \sqrt{\sum_I \texttt{src1}(I)^2} }{if  \(\texttt{normType} = \texttt{NORM_L2}\) }\f]
+
+- Absolute norm for two arrays
+\f[norm =  \forkthree{\|\texttt{src1}-\texttt{src2}\|_{L_{\infty}} =  \max _I | \texttt{src1} (I) -  \texttt{src2} (I)|}{if  \(\texttt{normType} = \texttt{NORM_INF}\) }
+{ \| \texttt{src1} - \texttt{src2} \| _{L_1} =  \sum _I | \texttt{src1} (I) -  \texttt{src2} (I)|}{if  \(\texttt{normType} = \texttt{NORM_L1}\) }
+{ \| \texttt{src1} - \texttt{src2} \| _{L_2} =  \sqrt{\sum_I (\texttt{src1}(I) - \texttt{src2}(I))^2} }{if  \(\texttt{normType} = \texttt{NORM_L2}\) }\f]
+
+- Relative norm for two arrays
+\f[norm =  \forkthree{\frac{\|\texttt{src1}-\texttt{src2}\|_{L_{\infty}}    }{\|\texttt{src2}\|_{L_{\infty}} }}{if  \(\texttt{normType} = \texttt{NORM_RELATIVE_INF}\) }
+{ \frac{\|\texttt{src1}-\texttt{src2}\|_{L_1} }{\|\texttt{src2}\|_{L_1}} }{if  \(\texttt{normType} = \texttt{NORM_RELATIVE_L1}\) }
+{ \frac{\|\texttt{src1}-\texttt{src2}\|_{L_2} }{\|\texttt{src2}\|_{L_2}} }{if  \(\texttt{normType} = \texttt{NORM_RELATIVE_L2}\) }\f]
+
+As example for one array consider the function \f$r(x)= \begin{pmatrix} x \\ 1-x \end{pmatrix}, x \in [-1;1]\f$.
+The \f$ L_{1}, L_{2} \f$ and \f$ L_{\infty} \f$ norm for the sample value \f$r(-1) = \begin{pmatrix} -1 \\ 2 \end{pmatrix}\f$
+is calculated as follows
+\f{align*}
+    \| r(-1) \|_{L_1} &= |-1| + |2| = 3 \\
+    \| r(-1) \|_{L_2} &= \sqrt{(-1)^{2} + (2)^{2}} = \sqrt{5} \\
+    \| r(-1) \|_{L_\infty} &= \max(|-1|,|2|) = 2
+\f}
+and for \f$r(0.5) = \begin{pmatrix} 0.5 \\ 0.5 \end{pmatrix}\f$ the calculation is
+\f{align*}
+    \| r(0.5) \|_{L_1} &= |0.5| + |0.5| = 1 \\
+    \| r(0.5) \|_{L_2} &= \sqrt{(0.5)^{2} + (0.5)^{2}} = \sqrt{0.5} \\
+    \| r(0.5) \|_{L_\infty} &= \max(|0.5|,|0.5|) = 0.5.
+\f}
+The following graphic shows all values for the three norm functions \f$\| r(x) \|_{L_1}, \| r(x) \|_{L_2}\f$ and \f$\| r(x) \|_{L_\infty}\f$.
+It is notable that the \f$ L_{1} \f$ norm forms the upper and the \f$ L_{\infty} \f$ norm forms the lower border for the example function \f$ r(x) \f$.
+![Graphs for the different norm functions from the above example](pics/NormTypes_OneArray_1-2-INF.png)
+ */
+enum NormTypes { NORM_INF       = 1,
+                 NORM_L1        = 2,
+                 NORM_L2        = 4,
+                 NORM_L2SQR     = 5,
+                 NORM_HAMMING   = 6,
+                 NORM_HAMMING2  = 7,
+                 NORM_TYPE_MASK = 7,
+                 NORM_RELATIVE  = 8, //!< flag
+                 NORM_MINMAX    = 32 //!< flag
+               };
+
+//! comparison types
+enum CmpTypes { CMP_EQ = 0, //!< src1 is equal to src2.
+                CMP_GT = 1, //!< src1 is greater than src2.
+                CMP_GE = 2, //!< src1 is greater than or equal to src2.
+                CMP_LT = 3, //!< src1 is less than src2.
+                CMP_LE = 4, //!< src1 is less than or equal to src2.
+                CMP_NE = 5  //!< src1 is unequal to src2.
+              };
+
+//! generalized matrix multiplication flags
+enum GemmFlags { GEMM_1_T = 1, //!< transposes src1
+                 GEMM_2_T = 2, //!< transposes src2
+                 GEMM_3_T = 4 //!< transposes src3
+               };
+
+enum DftFlags {
+    /** performs an inverse 1D or 2D transform instead of the default forward
+        transform. */
+    DFT_INVERSE        = 1,
+    /** scales the result: divide it by the number of array elements. Normally, it is
+        combined with DFT_INVERSE. */
+    DFT_SCALE          = 2,
+    /** performs a forward or inverse transform of every individual row of the input
+        matrix; this flag enables you to transform multiple vectors simultaneously and can be used to
+        decrease the overhead (which is sometimes several times larger than the processing itself) to
+        perform 3D and higher-dimensional transformations and so forth.*/
+    DFT_ROWS           = 4,
+    /** performs a forward transformation of 1D or 2D real array; the result,
+        though being a complex array, has complex-conjugate symmetry (*CCS*, see the function
+        description below for details), and such an array can be packed into a real array of the same
+        size as input, which is the fastest option and which is what the function does by default;
+        however, you may wish to get a full complex array (for simpler spectrum analysis, and so on) -
+        pass the flag to enable the function to produce a full-size complex output array. */
+    DFT_COMPLEX_OUTPUT = 16,
+    /** performs an inverse transformation of a 1D or 2D complex array; the
+        result is normally a complex array of the same size, however, if the input array has
+        conjugate-complex symmetry (for example, it is a result of forward transformation with
+        DFT_COMPLEX_OUTPUT flag), the output is a real array; while the function itself does not
+        check whether the input is symmetrical or not, you can pass the flag and then the function
+        will assume the symmetry and produce the real output array (note that when the input is packed
+        into a real array and inverse transformation is executed, the function treats the input as a
+        packed complex-conjugate symmetrical array, and the output will also be a real array). */
+    DFT_REAL_OUTPUT    = 32,
+    /** performs an inverse 1D or 2D transform instead of the default forward transform. */
+    DCT_INVERSE        = DFT_INVERSE,
+    /** performs a forward or inverse transform of every individual row of the input
+        matrix. This flag enables you to transform multiple vectors simultaneously and can be used to
+        decrease the overhead (which is sometimes several times larger than the processing itself) to
+        perform 3D and higher-dimensional transforms and so forth.*/
+    DCT_ROWS           = DFT_ROWS
+};
+
+//! Various border types, image boundaries are denoted with `|`
+//! @see borderInterpolate, copyMakeBorder
+enum BorderTypes {
+    BORDER_CONSTANT    = 0, //!< `iiiiii|abcdefgh|iiiiiii`  with some specified `i`
+    BORDER_REPLICATE   = 1, //!< `aaaaaa|abcdefgh|hhhhhhh`
+    BORDER_REFLECT     = 2, //!< `fedcba|abcdefgh|hgfedcb`
+    BORDER_WRAP        = 3, //!< `cdefgh|abcdefgh|abcdefg`
+    BORDER_REFLECT_101 = 4, //!< `gfedcb|abcdefgh|gfedcba`
+    BORDER_TRANSPARENT = 5, //!< `uvwxyz|absdefgh|ijklmno`
+
+    BORDER_REFLECT101  = BORDER_REFLECT_101, //!< same as BORDER_REFLECT_101
+    BORDER_DEFAULT     = BORDER_REFLECT_101, //!< same as BORDER_REFLECT_101
+    BORDER_ISOLATED    = 16 //!< do not look outside of ROI
+};
+
+//! @} core_array
+
+//! @addtogroup core_utils
+//! @{
+
+//! @cond IGNORED
+
+//////////////// static assert /////////////////
+#define CVAUX_CONCAT_EXP(a, b) a##b
+#define CVAUX_CONCAT(a, b) CVAUX_CONCAT_EXP(a,b)
+
+#if defined(__clang__)
+#  ifndef __has_extension
+#    define __has_extension __has_feature /* compatibility, for older versions of clang */
+#  endif
+#  if __has_extension(cxx_static_assert)
+#    define CV_StaticAssert(condition, reason)    static_assert((condition), reason " " #condition)
+#  elif __has_extension(c_static_assert)
+#    define CV_StaticAssert(condition, reason)    _Static_assert((condition), reason " " #condition)
+#  endif
+#elif defined(__GNUC__)
+#  if (defined(__GXX_EXPERIMENTAL_CXX0X__) || __cplusplus >= 201103L)
+#    define CV_StaticAssert(condition, reason)    static_assert((condition), reason " " #condition)
+#  endif
+#elif defined(_MSC_VER)
+#  if _MSC_VER >= 1600 /* MSVC 10 */
+#    define CV_StaticAssert(condition, reason)    static_assert((condition), reason " " #condition)
+#  endif
+#endif
+#ifndef CV_StaticAssert
+#  if !defined(__clang__) && defined(__GNUC__) && (__GNUC__*100 + __GNUC_MINOR__ > 302)
+#    define CV_StaticAssert(condition, reason) ({ extern int __attribute__((error("CV_StaticAssert: " reason " " #condition))) CV_StaticAssert(); ((condition) ? 0 : CV_StaticAssert()); })
+#  else
+     template <bool x> struct CV_StaticAssert_failed;
+     template <> struct CV_StaticAssert_failed<true> { enum { val = 1 }; };
+     template<int x> struct CV_StaticAssert_test {};
+#    define CV_StaticAssert(condition, reason)\
+       typedef cv::CV_StaticAssert_test< sizeof(cv::CV_StaticAssert_failed< static_cast<bool>(condition) >) > CVAUX_CONCAT(CV_StaticAssert_failed_at_, __LINE__)
+#  endif
+#endif
+
+// Suppress warning "-Wdeprecated-declarations" / C4996
+#if defined(_MSC_VER)
+    #define CV_DO_PRAGMA(x) __pragma(x)
+#elif defined(__GNUC__)
+    #define CV_DO_PRAGMA(x) _Pragma (#x)
+#else
+    #define CV_DO_PRAGMA(x)
+#endif
+
+#ifdef _MSC_VER
+#define CV_SUPPRESS_DEPRECATED_START \
+    CV_DO_PRAGMA(warning(push)) \
+    CV_DO_PRAGMA(warning(disable: 4996))
+#define CV_SUPPRESS_DEPRECATED_END CV_DO_PRAGMA(warning(pop))
+#elif defined (__clang__) || ((__GNUC__)  && (__GNUC__*100 + __GNUC_MINOR__ > 405))
+#define CV_SUPPRESS_DEPRECATED_START \
+    CV_DO_PRAGMA(GCC diagnostic push) \
+    CV_DO_PRAGMA(GCC diagnostic ignored "-Wdeprecated-declarations")
+#define CV_SUPPRESS_DEPRECATED_END CV_DO_PRAGMA(GCC diagnostic pop)
+#else
+#define CV_SUPPRESS_DEPRECATED_START
+#define CV_SUPPRESS_DEPRECATED_END
+#endif
+#define CV_UNUSED(name) (void)name
+//! @endcond
+
+/*! @brief Signals an error and raises the exception.
+
+By default the function prints information about the error to stderr,
+then it either stops if setBreakOnError() had been called before or raises the exception.
+It is possible to alternate error processing by using redirectError().
+@param _code - error code (Error::Code)
+@param _err - error description
+@param _func - function name. Available only when the compiler supports getting it
+@param _file - source file name where the error has occured
+@param _line - line number in the source file where the error has occured
+@see CV_Error, CV_Error_, CV_ErrorNoReturn, CV_ErrorNoReturn_, CV_Assert, CV_DbgAssert
+ */
+CV_EXPORTS void error(int _code, const String& _err, const char* _func, const char* _file, int _line);
+
+#ifdef __GNUC__
+# if defined __clang__ || defined __APPLE__
+#   pragma GCC diagnostic push
+#   pragma GCC diagnostic ignored "-Winvalid-noreturn"
+# endif
+#endif
+
+/** same as cv::error, but does not return */
+CV_INLINE CV_NORETURN void errorNoReturn(int _code, const String& _err, const char* _func, const char* _file, int _line)
+{
+    error(_code, _err, _func, _file, _line);
+#ifdef __GNUC__
+# if !defined __clang__ && !defined __APPLE__
+    // this suppresses this warning: "noreturn" function does return [enabled by default]
+    __builtin_trap();
+    // or use infinite loop: for (;;) {}
+# endif
+#endif
+}
+#ifdef __GNUC__
+# if defined __clang__ || defined __APPLE__
+#   pragma GCC diagnostic pop
+# endif
+#endif
+
+#if defined __GNUC__
+#define CV_Func __func__
+#elif defined _MSC_VER
+#define CV_Func __FUNCTION__
+#else
+#define CV_Func ""
+#endif
+
+/** @brief Call the error handler.
+
+Currently, the error handler prints the error code and the error message to the standard
+error stream `stderr`. In the Debug configuration, it then provokes memory access violation, so that
+the execution stack and all the parameters can be analyzed by the debugger. In the Release
+configuration, the exception is thrown.
+
+@param code one of Error::Code
+@param msg error message
+*/
+#define CV_Error( code, msg ) cv::error( code, msg, CV_Func, __FILE__, __LINE__ )
+
+/**  @brief Call the error handler.
+
+This macro can be used to construct an error message on-fly to include some dynamic information,
+for example:
+@code
+    // note the extra parentheses around the formatted text message
+    CV_Error_( CV_StsOutOfRange,
+    ("the value at (%d, %d)=%g is out of range", badPt.x, badPt.y, badValue));
+@endcode
+@param code one of Error::Code
+@param args printf-like formatted error message in parentheses
+*/
+#define CV_Error_( code, args ) cv::error( code, cv::format args, CV_Func, __FILE__, __LINE__ )
+
+/** @brief Checks a condition at runtime and throws exception if it fails
+
+The macros CV_Assert (and CV_DbgAssert(expr)) evaluate the specified expression. If it is 0, the macros
+raise an error (see cv::error). The macro CV_Assert checks the condition in both Debug and Release
+configurations while CV_DbgAssert is only retained in the Debug configuration.
+*/
+#define CV_Assert( expr ) if(!!(expr)) ; else cv::error( cv::Error::StsAssert, #expr, CV_Func, __FILE__, __LINE__ )
+
+/** same as CV_Error(code,msg), but does not return */
+#define CV_ErrorNoReturn( code, msg ) cv::errorNoReturn( code, msg, CV_Func, __FILE__, __LINE__ )
+
+/** same as CV_Error_(code,args), but does not return */
+#define CV_ErrorNoReturn_( code, args ) cv::errorNoReturn( code, cv::format args, CV_Func, __FILE__, __LINE__ )
+
+/** replaced with CV_Assert(expr) in Debug configuration */
+#ifdef _DEBUG
+#  define CV_DbgAssert(expr) CV_Assert(expr)
+#else
+#  define CV_DbgAssert(expr)
+#endif
+
+/*
+ * Hamming distance functor - counts the bit differences between two strings - useful for the Brief descriptor
+ * bit count of A exclusive XOR'ed with B
+ */
+struct CV_EXPORTS Hamming
+{
+    enum { normType = NORM_HAMMING };
+    typedef unsigned char ValueType;
+    typedef int ResultType;
+
+    /** this will count the bits in a ^ b
+     */
+    ResultType operator()( const unsigned char* a, const unsigned char* b, int size ) const;
+};
+
+typedef Hamming HammingLUT;
+
+/////////////////////////////////// inline norms ////////////////////////////////////
+
+template<typename _Tp> inline _Tp cv_abs(_Tp x) { return std::abs(x); }
+inline int cv_abs(uchar x) { return x; }
+inline int cv_abs(schar x) { return std::abs(x); }
+inline int cv_abs(ushort x) { return x; }
+inline int cv_abs(short x) { return std::abs(x); }
+
+template<typename _Tp, typename _AccTp> static inline
+_AccTp normL2Sqr(const _Tp* a, int n)
+{
+    _AccTp s = 0;
+    int i=0;
+#if CV_ENABLE_UNROLLED
+    for( ; i <= n - 4; i += 4 )
+    {
+        _AccTp v0 = a[i], v1 = a[i+1], v2 = a[i+2], v3 = a[i+3];
+        s += v0*v0 + v1*v1 + v2*v2 + v3*v3;
+    }
+#endif
+    for( ; i < n; i++ )
+    {
+        _AccTp v = a[i];
+        s += v*v;
+    }
+    return s;
+}
+
+template<typename _Tp, typename _AccTp> static inline
+_AccTp normL1(const _Tp* a, int n)
+{
+    _AccTp s = 0;
+    int i = 0;
+#if CV_ENABLE_UNROLLED
+    for(; i <= n - 4; i += 4 )
+    {
+        s += (_AccTp)cv_abs(a[i]) + (_AccTp)cv_abs(a[i+1]) +
+            (_AccTp)cv_abs(a[i+2]) + (_AccTp)cv_abs(a[i+3]);
+    }
+#endif
+    for( ; i < n; i++ )
+        s += cv_abs(a[i]);
+    return s;
+}
+
+template<typename _Tp, typename _AccTp> static inline
+_AccTp normInf(const _Tp* a, int n)
+{
+    _AccTp s = 0;
+    for( int i = 0; i < n; i++ )
+        s = std::max(s, (_AccTp)cv_abs(a[i]));
+    return s;
+}
+
+template<typename _Tp, typename _AccTp> static inline
+_AccTp normL2Sqr(const _Tp* a, const _Tp* b, int n)
+{
+    _AccTp s = 0;
+    int i= 0;
+#if CV_ENABLE_UNROLLED
+    for(; i <= n - 4; i += 4 )
+    {
+        _AccTp v0 = _AccTp(a[i] - b[i]), v1 = _AccTp(a[i+1] - b[i+1]), v2 = _AccTp(a[i+2] - b[i+2]), v3 = _AccTp(a[i+3] - b[i+3]);
+        s += v0*v0 + v1*v1 + v2*v2 + v3*v3;
+    }
+#endif
+    for( ; i < n; i++ )
+    {
+        _AccTp v = _AccTp(a[i] - b[i]);
+        s += v*v;
+    }
+    return s;
+}
+
+static inline float normL2Sqr(const float* a, const float* b, int n)
+{
+    float s = 0.f;
+    for( int i = 0; i < n; i++ )
+    {
+        float v = a[i] - b[i];
+        s += v*v;
+    }
+    return s;
+}
+
+template<typename _Tp, typename _AccTp> static inline
+_AccTp normL1(const _Tp* a, const _Tp* b, int n)
+{
+    _AccTp s = 0;
+    int i= 0;
+#if CV_ENABLE_UNROLLED
+    for(; i <= n - 4; i += 4 )
+    {
+        _AccTp v0 = _AccTp(a[i] - b[i]), v1 = _AccTp(a[i+1] - b[i+1]), v2 = _AccTp(a[i+2] - b[i+2]), v3 = _AccTp(a[i+3] - b[i+3]);
+        s += std::abs(v0) + std::abs(v1) + std::abs(v2) + std::abs(v3);
+    }
+#endif
+    for( ; i < n; i++ )
+    {
+        _AccTp v = _AccTp(a[i] - b[i]);
+        s += std::abs(v);
+    }
+    return s;
+}
+
+inline float normL1(const float* a, const float* b, int n)
+{
+    float s = 0.f;
+    for( int i = 0; i < n; i++ )
+    {
+        s += std::abs(a[i] - b[i]);
+    }
+    return s;
+}
+
+inline int normL1(const uchar* a, const uchar* b, int n)
+{
+    int s = 0;
+    for( int i = 0; i < n; i++ )
+    {
+        s += std::abs(a[i] - b[i]);
+    }
+    return s;
+}
+
+template<typename _Tp, typename _AccTp> static inline
+_AccTp normInf(const _Tp* a, const _Tp* b, int n)
+{
+    _AccTp s = 0;
+    for( int i = 0; i < n; i++ )
+    {
+        _AccTp v0 = a[i] - b[i];
+        s = std::max(s, std::abs(v0));
+    }
+    return s;
+}
+
+/** @brief Computes the cube root of an argument.
+
+ The function cubeRoot computes \f$\sqrt[3]{\texttt{val}}\f$. Negative arguments are handled correctly.
+ NaN and Inf are not handled. The accuracy approaches the maximum possible accuracy for
+ single-precision data.
+ @param val A function argument.
+ */
+CV_EXPORTS_W float cubeRoot(float val);
+
+/** @brief Calculates the angle of a 2D vector in degrees.
+
+ The function fastAtan2 calculates the full-range angle of an input 2D vector. The angle is measured
+ in degrees and varies from 0 to 360 degrees. The accuracy is about 0.3 degrees.
+ @param x x-coordinate of the vector.
+ @param y y-coordinate of the vector.
+ */
+CV_EXPORTS_W float fastAtan2(float y, float x);
+
+/** proxy for hal::LU */
+CV_EXPORTS int LU(float* A, size_t astep, int m, float* b, size_t bstep, int n);
+/** proxy for hal::LU */
+CV_EXPORTS int LU(double* A, size_t astep, int m, double* b, size_t bstep, int n);
+/** proxy for hal::Cholesky */
+CV_EXPORTS bool Cholesky(float* A, size_t astep, int m, float* b, size_t bstep, int n);
+/** proxy for hal::Cholesky */
+CV_EXPORTS bool Cholesky(double* A, size_t astep, int m, double* b, size_t bstep, int n);
+
+////////////////// forward declarations for important OpenCV types //////////////////
+
+//! @cond IGNORED
+
+template<typename _Tp, int cn> class Vec;
+template<typename _Tp, int m, int n> class Matx;
+
+template<typename _Tp> class Complex;
+template<typename _Tp> class Point_;
+template<typename _Tp> class Point3_;
+template<typename _Tp> class Size_;
+template<typename _Tp> class Rect_;
+template<typename _Tp> class Scalar_;
+
+class CV_EXPORTS RotatedRect;
+class CV_EXPORTS Range;
+class CV_EXPORTS TermCriteria;
+class CV_EXPORTS KeyPoint;
+class CV_EXPORTS DMatch;
+class CV_EXPORTS RNG;
+
+class CV_EXPORTS Mat;
+class CV_EXPORTS MatExpr;
+
+class CV_EXPORTS UMat;
+
+class CV_EXPORTS SparseMat;
+typedef Mat MatND;
+
+template<typename _Tp> class Mat_;
+template<typename _Tp> class SparseMat_;
+
+class CV_EXPORTS MatConstIterator;
+class CV_EXPORTS SparseMatIterator;
+class CV_EXPORTS SparseMatConstIterator;
+template<typename _Tp> class MatIterator_;
+template<typename _Tp> class MatConstIterator_;
+template<typename _Tp> class SparseMatIterator_;
+template<typename _Tp> class SparseMatConstIterator_;
+
+namespace ogl
+{
+    class CV_EXPORTS Buffer;
+    class CV_EXPORTS Texture2D;
+    class CV_EXPORTS Arrays;
+}
+
+namespace cuda
+{
+    class CV_EXPORTS GpuMat;
+    class CV_EXPORTS HostMem;
+    class CV_EXPORTS Stream;
+    class CV_EXPORTS Event;
+}
+
+namespace cudev
+{
+    template <typename _Tp> class GpuMat_;
+}
+
+namespace ipp
+{
+CV_EXPORTS int getIppFeatures();
+CV_EXPORTS void setIppStatus(int status, const char * const funcname = NULL, const char * const filename = NULL,
+                             int line = 0);
+CV_EXPORTS int getIppStatus();
+CV_EXPORTS String getIppErrorLocation();
+CV_EXPORTS bool useIPP();
+CV_EXPORTS void setUseIPP(bool flag);
+
+} // ipp
+
+//! @endcond
+
+//! @} core_utils
+
+
+
+
+} // cv
+
+#include "opencv2/core/neon_utils.hpp"
+
+#endif //OPENCV_CORE_BASE_HPP

include/opencv2/core/bufferpool.hpp

@@ -0,0 +1,31 @@
+// This file is part of OpenCV project.
+// It is subject to the license terms in the LICENSE file found in the top-level directory
+// of this distribution and at http://opencv.org/license.html.
+//
+// Copyright (C) 2014, Advanced Micro Devices, Inc., all rights reserved.
+
+#ifndef OPENCV_CORE_BUFFER_POOL_HPP
+#define OPENCV_CORE_BUFFER_POOL_HPP
+
+namespace cv
+{
+
+//! @addtogroup core
+//! @{
+
+class BufferPoolController
+{
+protected:
+    ~BufferPoolController() { }
+public:
+    virtual size_t getReservedSize() const = 0;
+    virtual size_t getMaxReservedSize() const = 0;
+    virtual void setMaxReservedSize(size_t size) = 0;
+    virtual void freeAllReservedBuffers() = 0;
+};
+
+//! @}
+
+}
+
+#endif // OPENCV_CORE_BUFFER_POOL_HPP

include/opencv2/core/cuda.hpp

@@ -0,0 +1,874 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                          License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef OPENCV_CORE_CUDA_HPP
+#define OPENCV_CORE_CUDA_HPP
+
+#ifndef __cplusplus
+#  error cuda.hpp header must be compiled as C++
+#endif
+
+#include "opencv2/core.hpp"
+#include "opencv2/core/cuda_types.hpp"
+
+/**
+  @defgroup cuda CUDA-accelerated Computer Vision
+  @{
+    @defgroup cudacore Core part
+    @{
+      @defgroup cudacore_init Initalization and Information
+      @defgroup cudacore_struct Data Structures
+    @}
+  @}
+ */
+
+namespace cv { namespace cuda {
+
+//! @addtogroup cudacore_struct
+//! @{
+
+//===================================================================================
+// GpuMat
+//===================================================================================
+
+/** @brief Base storage class for GPU memory with reference counting.
+
+Its interface matches the Mat interface with the following limitations:
+
+-   no arbitrary dimensions support (only 2D)
+-   no functions that return references to their data (because references on GPU are not valid for
+    CPU)
+-   no expression templates technique support
+
+Beware that the latter limitation may lead to overloaded matrix operators that cause memory
+allocations. The GpuMat class is convertible to cuda::PtrStepSz and cuda::PtrStep so it can be
+passed directly to the kernel.
+
+@note In contrast with Mat, in most cases GpuMat::isContinuous() == false . This means that rows are
+aligned to a size depending on the hardware. Single-row GpuMat is always a continuous matrix.
+
+@note You are not recommended to leave static or global GpuMat variables allocated, that is, to rely
+on its destructor. The destruction order of such variables and CUDA context is undefined. GPU memory
+release function returns error if the CUDA context has been destroyed before.
+
+@sa Mat
+ */
+class CV_EXPORTS GpuMat
+{
+public:
+    class CV_EXPORTS Allocator
+    {
+    public:
+        virtual ~Allocator() {}
+
+        // allocator must fill data, step and refcount fields
+        virtual bool allocate(GpuMat* mat, int rows, int cols, size_t elemSize) = 0;
+        virtual void free(GpuMat* mat) = 0;
+    };
+
+    //! default allocator
+    static Allocator* defaultAllocator();
+    static void setDefaultAllocator(Allocator* allocator);
+
+    //! default constructor
+    explicit GpuMat(Allocator* allocator = defaultAllocator());
+
+    //! constructs GpuMat of the specified size and type
+    GpuMat(int rows, int cols, int type, Allocator* allocator = defaultAllocator());
+    GpuMat(Size size, int type, Allocator* allocator = defaultAllocator());
+
+    //! constucts GpuMat and fills it with the specified value _s
+    GpuMat(int rows, int cols, int type, Scalar s, Allocator* allocator = defaultAllocator());
+    GpuMat(Size size, int type, Scalar s, Allocator* allocator = defaultAllocator());
+
+    //! copy constructor
+    GpuMat(const GpuMat& m);
+
+    //! constructor for GpuMat headers pointing to user-allocated data
+    GpuMat(int rows, int cols, int type, void* data, size_t step = Mat::AUTO_STEP);
+    GpuMat(Size size, int type, void* data, size_t step = Mat::AUTO_STEP);
+
+    //! creates a GpuMat header for a part of the bigger matrix
+    GpuMat(const GpuMat& m, Range rowRange, Range colRange);
+    GpuMat(const GpuMat& m, Rect roi);
+
+    //! builds GpuMat from host memory (Blocking call)
+    explicit GpuMat(InputArray arr, Allocator* allocator = defaultAllocator());
+
+    //! destructor - calls release()
+    ~GpuMat();
+
+    //! assignment operators
+    GpuMat& operator =(const GpuMat& m);
+
+    //! allocates new GpuMat data unless the GpuMat already has specified size and type
+    void create(int rows, int cols, int type);
+    void create(Size size, int type);
+
+    //! decreases reference counter, deallocate the data when reference counter reaches 0
+    void release();
+
+    //! swaps with other smart pointer
+    void swap(GpuMat& mat);
+
+    //! pefroms upload data to GpuMat (Blocking call)
+    void upload(InputArray arr);
+
+    //! pefroms upload data to GpuMat (Non-Blocking call)
+    void upload(InputArray arr, Stream& stream);
+
+    //! pefroms download data from device to host memory (Blocking call)
+    void download(OutputArray dst) const;
+
+    //! pefroms download data from device to host memory (Non-Blocking call)
+    void download(OutputArray dst, Stream& stream) const;
+
+    //! returns deep copy of the GpuMat, i.e. the data is copied
+    GpuMat clone() const;
+
+    //! copies the GpuMat content to device memory (Blocking call)
+    void copyTo(OutputArray dst) const;
+
+    //! copies the GpuMat content to device memory (Non-Blocking call)
+    void copyTo(OutputArray dst, Stream& stream) const;
+
+    //! copies those GpuMat elements to "m" that are marked with non-zero mask elements (Blocking call)
+    void copyTo(OutputArray dst, InputArray mask) const;
+
+    //! copies those GpuMat elements to "m" that are marked with non-zero mask elements (Non-Blocking call)
+    void copyTo(OutputArray dst, InputArray mask, Stream& stream) const;
+
+    //! sets some of the GpuMat elements to s (Blocking call)
+    GpuMat& setTo(Scalar s);
+
+    //! sets some of the GpuMat elements to s (Non-Blocking call)
+    GpuMat& setTo(Scalar s, Stream& stream);
+
+    //! sets some of the GpuMat elements to s, according to the mask (Blocking call)
+    GpuMat& setTo(Scalar s, InputArray mask);
+
+    //! sets some of the GpuMat elements to s, according to the mask (Non-Blocking call)
+    GpuMat& setTo(Scalar s, InputArray mask, Stream& stream);
+
+    //! converts GpuMat to another datatype (Blocking call)
+    void convertTo(OutputArray dst, int rtype) const;
+
+    //! converts GpuMat to another datatype (Non-Blocking call)
+    void convertTo(OutputArray dst, int rtype, Stream& stream) const;
+
+    //! converts GpuMat to another datatype with scaling (Blocking call)
+    void convertTo(OutputArray dst, int rtype, double alpha, double beta = 0.0) const;
+
+    //! converts GpuMat to another datatype with scaling (Non-Blocking call)
+    void convertTo(OutputArray dst, int rtype, double alpha, Stream& stream) const;
+
+    //! converts GpuMat to another datatype with scaling (Non-Blocking call)
+    void convertTo(OutputArray dst, int rtype, double alpha, double beta, Stream& stream) const;
+
+    void assignTo(GpuMat& m, int type=-1) const;
+
+    //! returns pointer to y-th row
+    uchar* ptr(int y = 0);
+    const uchar* ptr(int y = 0) const;
+
+    //! template version of the above method
+    template<typename _Tp> _Tp* ptr(int y = 0);
+    template<typename _Tp> const _Tp* ptr(int y = 0) const;
+
+    template <typename _Tp> operator PtrStepSz<_Tp>() const;
+    template <typename _Tp> operator PtrStep<_Tp>() const;
+
+    //! returns a new GpuMat header for the specified row
+    GpuMat row(int y) const;
+
+    //! returns a new GpuMat header for the specified column
+    GpuMat col(int x) const;
+
+    //! ... for the specified row span
+    GpuMat rowRange(int startrow, int endrow) const;
+    GpuMat rowRange(Range r) const;
+
+    //! ... for the specified column span
+    GpuMat colRange(int startcol, int endcol) const;
+    GpuMat colRange(Range r) const;
+
+    //! extracts a rectangular sub-GpuMat (this is a generalized form of row, rowRange etc.)
+    GpuMat operator ()(Range rowRange, Range colRange) const;
+    GpuMat operator ()(Rect roi) const;
+
+    //! creates alternative GpuMat header for the same data, with different
+    //! number of channels and/or different number of rows
+    GpuMat reshape(int cn, int rows = 0) const;
+
+    //! locates GpuMat header within a parent GpuMat
+    void locateROI(Size& wholeSize, Point& ofs) const;
+
+    //! moves/resizes the current GpuMat ROI inside the parent GpuMat
+    GpuMat& adjustROI(int dtop, int dbottom, int dleft, int dright);
+
+    //! returns true iff the GpuMat data is continuous
+    //! (i.e. when there are no gaps between successive rows)
+    bool isContinuous() const;
+
+    //! returns element size in bytes
+    size_t elemSize() const;
+
+    //! returns the size of element channel in bytes
+    size_t elemSize1() const;
+
+    //! returns element type
+    int type() const;
+
+    //! returns element type
+    int depth() const;
+
+    //! returns number of channels
+    int channels() const;
+
+    //! returns step/elemSize1()
+    size_t step1() const;
+
+    //! returns GpuMat size : width == number of columns, height == number of rows
+    Size size() const;
+
+    //! returns true if GpuMat data is NULL
+    bool empty() const;
+
+    /*! includes several bit-fields:
+    - the magic signature
+    - continuity flag
+    - depth
+    - number of channels
+    */
+    int flags;
+
+    //! the number of rows and columns
+    int rows, cols;
+
+    //! a distance between successive rows in bytes; includes the gap if any
+    size_t step;
+
+    //! pointer to the data
+    uchar* data;
+
+    //! pointer to the reference counter;
+    //! when GpuMat points to user-allocated data, the pointer is NULL
+    int* refcount;
+
+    //! helper fields used in locateROI and adjustROI
+    uchar* datastart;
+    const uchar* dataend;
+
+    //! allocator
+    Allocator* allocator;
+};
+
+/** @brief Creates a continuous matrix.
+
+@param rows Row count.
+@param cols Column count.
+@param type Type of the matrix.
+@param arr Destination matrix. This parameter changes only if it has a proper type and area (
+\f$\texttt{rows} \times \texttt{cols}\f$ ).
+
+Matrix is called continuous if its elements are stored continuously, that is, without gaps at the
+end of each row.
+ */
+CV_EXPORTS void createContinuous(int rows, int cols, int type, OutputArray arr);
+
+/** @brief Ensures that the size of a matrix is big enough and the matrix has a proper type.
+
+@param rows Minimum desired number of rows.
+@param cols Minimum desired number of columns.
+@param type Desired matrix type.
+@param arr Destination matrix.
+
+The function does not reallocate memory if the matrix has proper attributes already.
+ */
+CV_EXPORTS void ensureSizeIsEnough(int rows, int cols, int type, OutputArray arr);
+
+//! BufferPool management (must be called before Stream creation)
+CV_EXPORTS void setBufferPoolUsage(bool on);
+CV_EXPORTS void setBufferPoolConfig(int deviceId, size_t stackSize, int stackCount);
+
+//===================================================================================
+// HostMem
+//===================================================================================
+
+/** @brief Class with reference counting wrapping special memory type allocation functions from CUDA.
+
+Its interface is also Mat-like but with additional memory type parameters.
+
+-   **PAGE_LOCKED** sets a page locked memory type used commonly for fast and asynchronous
+    uploading/downloading data from/to GPU.
+-   **SHARED** specifies a zero copy memory allocation that enables mapping the host memory to GPU
+    address space, if supported.
+-   **WRITE_COMBINED** sets the write combined buffer that is not cached by CPU. Such buffers are
+    used to supply GPU with data when GPU only reads it. The advantage is a better CPU cache
+    utilization.
+
+@note Allocation size of such memory types is usually limited. For more details, see *CUDA 2.2
+Pinned Memory APIs* document or *CUDA C Programming Guide*.
+ */
+class CV_EXPORTS HostMem
+{
+public:
+    enum AllocType { PAGE_LOCKED = 1, SHARED = 2, WRITE_COMBINED = 4 };
+
+    static MatAllocator* getAllocator(AllocType alloc_type = PAGE_LOCKED);
+
+    explicit HostMem(AllocType alloc_type = PAGE_LOCKED);
+
+    HostMem(const HostMem& m);
+
+    HostMem(int rows, int cols, int type, AllocType alloc_type = PAGE_LOCKED);
+    HostMem(Size size, int type, AllocType alloc_type = PAGE_LOCKED);
+
+    //! creates from host memory with coping data
+    explicit HostMem(InputArray arr, AllocType alloc_type = PAGE_LOCKED);
+
+    ~HostMem();
+
+    HostMem& operator =(const HostMem& m);
+
+    //! swaps with other smart pointer
+    void swap(HostMem& b);
+
+    //! returns deep copy of the matrix, i.e. the data is copied
+    HostMem clone() const;
+
+    //! allocates new matrix data unless the matrix already has specified size and type.
+    void create(int rows, int cols, int type);
+    void create(Size size, int type);
+
+    //! creates alternative HostMem header for the same data, with different
+    //! number of channels and/or different number of rows
+    HostMem reshape(int cn, int rows = 0) const;
+
+    //! decrements reference counter and released memory if needed.
+    void release();
+
+    //! returns matrix header with disabled reference counting for HostMem data.
+    Mat createMatHeader() const;
+
+    /** @brief Maps CPU memory to GPU address space and creates the cuda::GpuMat header without reference counting
+    for it.
+
+    This can be done only if memory was allocated with the SHARED flag and if it is supported by the
+    hardware. Laptops often share video and CPU memory, so address spaces can be mapped, which
+    eliminates an extra copy.
+     */
+    GpuMat createGpuMatHeader() const;
+
+    // Please see cv::Mat for descriptions
+    bool isContinuous() const;
+    size_t elemSize() const;
+    size_t elemSize1() const;
+    int type() const;
+    int depth() const;
+    int channels() const;
+    size_t step1() const;
+    Size size() const;
+    bool empty() const;
+
+    // Please see cv::Mat for descriptions
+    int flags;
+    int rows, cols;
+    size_t step;
+
+    uchar* data;
+    int* refcount;
+
+    uchar* datastart;
+    const uchar* dataend;
+
+    AllocType alloc_type;
+};
+
+/** @brief Page-locks the memory of matrix and maps it for the device(s).
+
+@param m Input matrix.
+ */
+CV_EXPORTS void registerPageLocked(Mat& m);
+
+/** @brief Unmaps the memory of matrix and makes it pageable again.
+
+@param m Input matrix.
+ */
+CV_EXPORTS void unregisterPageLocked(Mat& m);
+
+//===================================================================================
+// Stream
+//===================================================================================
+
+/** @brief This class encapsulates a queue of asynchronous calls.
+
+@note Currently, you may face problems if an operation is enqueued twice with different data. Some
+functions use the constant GPU memory, and next call may update the memory before the previous one
+has been finished. But calling different operations asynchronously is safe because each operation
+has its own constant buffer. Memory copy/upload/download/set operations to the buffers you hold are
+also safe.
+
+@note The Stream class is not thread-safe. Please use different Stream objects for different CPU threads.
+
+@code
+void thread1()
+{
+    cv::cuda::Stream stream1;
+    cv::cuda::func1(..., stream1);
+}
+
+void thread2()
+{
+    cv::cuda::Stream stream2;
+    cv::cuda::func2(..., stream2);
+}
+@endcode
+
+@note By default all CUDA routines are launched in Stream::Null() object, if the stream is not specified by user.
+In multi-threading environment the stream objects must be passed explicitly (see previous note).
+ */
+class CV_EXPORTS Stream
+{
+    typedef void (Stream::*bool_type)() const;
+    void this_type_does_not_support_comparisons() const {}
+
+public:
+    typedef void (*StreamCallback)(int status, void* userData);
+
+    //! creates a new asynchronous stream
+    Stream();
+
+    /** @brief Returns true if the current stream queue is finished. Otherwise, it returns false.
+    */
+    bool queryIfComplete() const;
+
+    /** @brief Blocks the current CPU thread until all operations in the stream are complete.
+    */
+    void waitForCompletion();
+
+    /** @brief Makes a compute stream wait on an event.
+    */
+    void waitEvent(const Event& event);
+
+    /** @brief Adds a callback to be called on the host after all currently enqueued items in the stream have
+    completed.
+
+    @note Callbacks must not make any CUDA API calls. Callbacks must not perform any synchronization
+    that may depend on outstanding device work or other callbacks that are not mandated to run earlier.
+    Callbacks without a mandated order (in independent streams) execute in undefined order and may be
+    serialized.
+     */
+    void enqueueHostCallback(StreamCallback callback, void* userData);
+
+    //! return Stream object for default CUDA stream
+    static Stream& Null();
+
+    //! returns true if stream object is not default (!= 0)
+    operator bool_type() const;
+
+    class Impl;
+
+private:
+    Ptr<Impl> impl_;
+    Stream(const Ptr<Impl>& impl);
+
+    friend struct StreamAccessor;
+    friend class BufferPool;
+    friend class DefaultDeviceInitializer;
+};
+
+class CV_EXPORTS Event
+{
+public:
+    enum CreateFlags
+    {
+        DEFAULT        = 0x00,  /**< Default event flag */
+        BLOCKING_SYNC  = 0x01,  /**< Event uses blocking synchronization */
+        DISABLE_TIMING = 0x02,  /**< Event will not record timing data */
+        INTERPROCESS   = 0x04   /**< Event is suitable for interprocess use. DisableTiming must be set */
+    };
+
+    explicit Event(CreateFlags flags = DEFAULT);
+
+    //! records an event
+    void record(Stream& stream = Stream::Null());
+
+    //! queries an event's status
+    bool queryIfComplete() const;
+
+    //! waits for an event to complete
+    void waitForCompletion();
+
+    //! computes the elapsed time between events
+    static float elapsedTime(const Event& start, const Event& end);
+
+    class Impl;
+
+private:
+    Ptr<Impl> impl_;
+    Event(const Ptr<Impl>& impl);
+
+    friend struct EventAccessor;
+};
+
+//! @} cudacore_struct
+
+//===================================================================================
+// Initialization & Info
+//===================================================================================
+
+//! @addtogroup cudacore_init
+//! @{
+
+/** @brief Returns the number of installed CUDA-enabled devices.
+
+Use this function before any other CUDA functions calls. If OpenCV is compiled without CUDA support,
+this function returns 0.
+ */
+CV_EXPORTS int getCudaEnabledDeviceCount();
+
+/** @brief Sets a device and initializes it for the current thread.
+
+@param device System index of a CUDA device starting with 0.
+
+If the call of this function is omitted, a default device is initialized at the fist CUDA usage.
+ */
+CV_EXPORTS void setDevice(int device);
+
+/** @brief Returns the current device index set by cuda::setDevice or initialized by default.
+ */
+CV_EXPORTS int getDevice();
+
+/** @brief Explicitly destroys and cleans up all resources associated with the current device in the current
+process.
+
+Any subsequent API call to this device will reinitialize the device.
+ */
+CV_EXPORTS void resetDevice();
+
+/** @brief Enumeration providing CUDA computing features.
+ */
+enum FeatureSet
+{
+    FEATURE_SET_COMPUTE_10 = 10,
+    FEATURE_SET_COMPUTE_11 = 11,
+    FEATURE_SET_COMPUTE_12 = 12,
+    FEATURE_SET_COMPUTE_13 = 13,
+    FEATURE_SET_COMPUTE_20 = 20,
+    FEATURE_SET_COMPUTE_21 = 21,
+    FEATURE_SET_COMPUTE_30 = 30,
+    FEATURE_SET_COMPUTE_32 = 32,
+    FEATURE_SET_COMPUTE_35 = 35,
+    FEATURE_SET_COMPUTE_50 = 50,
+
+    GLOBAL_ATOMICS = FEATURE_SET_COMPUTE_11,
+    SHARED_ATOMICS = FEATURE_SET_COMPUTE_12,
+    NATIVE_DOUBLE = FEATURE_SET_COMPUTE_13,
+    WARP_SHUFFLE_FUNCTIONS = FEATURE_SET_COMPUTE_30,
+    DYNAMIC_PARALLELISM = FEATURE_SET_COMPUTE_35
+};
+
+//! checks whether current device supports the given feature
+CV_EXPORTS bool deviceSupports(FeatureSet feature_set);
+
+/** @brief Class providing a set of static methods to check what NVIDIA\* card architecture the CUDA module was
+built for.
+
+According to the CUDA C Programming Guide Version 3.2: "PTX code produced for some specific compute
+capability can always be compiled to binary code of greater or equal compute capability".
+ */
+class CV_EXPORTS TargetArchs
+{
+public:
+    /** @brief The following method checks whether the module was built with the support of the given feature:
+
+    @param feature_set Features to be checked. See :ocvcuda::FeatureSet.
+     */
+    static bool builtWith(FeatureSet feature_set);
+
+    /** @brief There is a set of methods to check whether the module contains intermediate (PTX) or binary CUDA
+    code for the given architecture(s):
+
+    @param major Major compute capability version.
+    @param minor Minor compute capability version.
+     */
+    static bool has(int major, int minor);
+    static bool hasPtx(int major, int minor);
+    static bool hasBin(int major, int minor);
+
+    static bool hasEqualOrLessPtx(int major, int minor);
+    static bool hasEqualOrGreater(int major, int minor);
+    static bool hasEqualOrGreaterPtx(int major, int minor);
+    static bool hasEqualOrGreaterBin(int major, int minor);
+};
+
+/** @brief Class providing functionality for querying the specified GPU properties.
+ */
+class CV_EXPORTS DeviceInfo
+{
+public:
+    //! creates DeviceInfo object for the current GPU
+    DeviceInfo();
+
+    /** @brief The constructors.
+
+    @param device_id System index of the CUDA device starting with 0.
+
+    Constructs the DeviceInfo object for the specified device. If device_id parameter is missed, it
+    constructs an object for the current device.
+     */
+    DeviceInfo(int device_id);
+
+    /** @brief Returns system index of the CUDA device starting with 0.
+    */
+    int deviceID() const;
+
+    //! ASCII string identifying device
+    const char* name() const;
+
+    //! global memory available on device in bytes
+    size_t totalGlobalMem() const;
+
+    //! shared memory available per block in bytes
+    size_t sharedMemPerBlock() const;
+
+    //! 32-bit registers available per block
+    int regsPerBlock() const;
+
+    //! warp size in threads
+    int warpSize() const;
+
+    //! maximum pitch in bytes allowed by memory copies
+    size_t memPitch() const;
+
+    //! maximum number of threads per block
+    int maxThreadsPerBlock() const;
+
+    //! maximum size of each dimension of a block
+    Vec3i maxThreadsDim() const;
+
+    //! maximum size of each dimension of a grid
+    Vec3i maxGridSize() const;
+
+    //! clock frequency in kilohertz
+    int clockRate() const;
+
+    //! constant memory available on device in bytes
+    size_t totalConstMem() const;
+
+    //! major compute capability
+    int majorVersion() const;
+
+    //! minor compute capability
+    int minorVersion() const;
+
+    //! alignment requirement for textures
+    size_t textureAlignment() const;
+
+    //! pitch alignment requirement for texture references bound to pitched memory
+    size_t texturePitchAlignment() const;
+
+    //! number of multiprocessors on device
+    int multiProcessorCount() const;
+
+    //! specified whether there is a run time limit on kernels
+    bool kernelExecTimeoutEnabled() const;
+
+    //! device is integrated as opposed to discrete
+    bool integrated() const;
+
+    //! device can map host memory with cudaHostAlloc/cudaHostGetDevicePointer
+    bool canMapHostMemory() const;
+
+    enum ComputeMode
+    {
+        ComputeModeDefault,         /**< default compute mode (Multiple threads can use cudaSetDevice with this device) */
+        ComputeModeExclusive,       /**< compute-exclusive-thread mode (Only one thread in one process will be able to use cudaSetDevice with this device) */
+        ComputeModeProhibited,      /**< compute-prohibited mode (No threads can use cudaSetDevice with this device) */
+        ComputeModeExclusiveProcess /**< compute-exclusive-process mode (Many threads in one process will be able to use cudaSetDevice with this device) */
+    };
+
+    //! compute mode
+    ComputeMode computeMode() const;
+
+    //! maximum 1D texture size
+    int maxTexture1D() const;
+
+    //! maximum 1D mipmapped texture size
+    int maxTexture1DMipmap() const;
+
+    //! maximum size for 1D textures bound to linear memory
+    int maxTexture1DLinear() const;
+
+    //! maximum 2D texture dimensions
+    Vec2i maxTexture2D() const;
+
+    //! maximum 2D mipmapped texture dimensions
+    Vec2i maxTexture2DMipmap() const;
+
+    //! maximum dimensions (width, height, pitch) for 2D textures bound to pitched memory
+    Vec3i maxTexture2DLinear() const;
+
+    //! maximum 2D texture dimensions if texture gather operations have to be performed
+    Vec2i maxTexture2DGather() const;
+
+    //! maximum 3D texture dimensions
+    Vec3i maxTexture3D() const;
+
+    //! maximum Cubemap texture dimensions
+    int maxTextureCubemap() const;
+
+    //! maximum 1D layered texture dimensions
+    Vec2i maxTexture1DLayered() const;
+
+    //! maximum 2D layered texture dimensions
+    Vec3i maxTexture2DLayered() const;
+
+    //! maximum Cubemap layered texture dimensions
+    Vec2i maxTextureCubemapLayered() const;
+
+    //! maximum 1D surface size
+    int maxSurface1D() const;
+
+    //! maximum 2D surface dimensions
+    Vec2i maxSurface2D() const;
+
+    //! maximum 3D surface dimensions
+    Vec3i maxSurface3D() const;
+
+    //! maximum 1D layered surface dimensions
+    Vec2i maxSurface1DLayered() const;
+
+    //! maximum 2D layered surface dimensions
+    Vec3i maxSurface2DLayered() const;
+
+    //! maximum Cubemap surface dimensions
+    int maxSurfaceCubemap() const;
+
+    //! maximum Cubemap layered surface dimensions
+    Vec2i maxSurfaceCubemapLayered() const;
+
+    //! alignment requirements for surfaces
+    size_t surfaceAlignment() const;
+
+    //! device can possibly execute multiple kernels concurrently
+    bool concurrentKernels() const;
+
+    //! device has ECC support enabled
+    bool ECCEnabled() const;
+
+    //! PCI bus ID of the device
+    int pciBusID() const;
+
+    //! PCI device ID of the device
+    int pciDeviceID() const;
+
+    //! PCI domain ID of the device
+    int pciDomainID() const;
+
+    //! true if device is a Tesla device using TCC driver, false otherwise
+    bool tccDriver() const;
+
+    //! number of asynchronous engines
+    int asyncEngineCount() const;
+
+    //! device shares a unified address space with the host
+    bool unifiedAddressing() const;
+
+    //! peak memory clock frequency in kilohertz
+    int memoryClockRate() const;
+
+    //! global memory bus width in bits
+    int memoryBusWidth() const;
+
+    //! size of L2 cache in bytes
+    int l2CacheSize() const;
+
+    //! maximum resident threads per multiprocessor
+    int maxThreadsPerMultiProcessor() const;
+
+    //! gets free and total device memory
+    void queryMemory(size_t& totalMemory, size_t& freeMemory) const;
+    size_t freeMemory() const;
+    size_t totalMemory() const;
+
+    /** @brief Provides information on CUDA feature support.
+
+    @param feature_set Features to be checked. See cuda::FeatureSet.
+
+    This function returns true if the device has the specified CUDA feature. Otherwise, it returns false
+     */
+    bool supports(FeatureSet feature_set) const;
+
+    /** @brief Checks the CUDA module and device compatibility.
+
+    This function returns true if the CUDA module can be run on the specified device. Otherwise, it
+    returns false .
+     */
+    bool isCompatible() const;
+
+private:
+    int device_id_;
+};
+
+CV_EXPORTS void printCudaDeviceInfo(int device);
+CV_EXPORTS void printShortCudaDeviceInfo(int device);
+
+/** @brief Converts an array to half precision floating number.
+
+@param _src input array.
+@param _dst output array.
+@param stream Stream for the asynchronous version.
+@sa convertFp16
+*/
+CV_EXPORTS void convertFp16(InputArray _src, OutputArray _dst, Stream& stream = Stream::Null());
+
+//! @} cudacore_init
+
+}} // namespace cv { namespace cuda {
+
+
+#include "opencv2/core/cuda.inl.hpp"
+
+#endif /* OPENCV_CORE_CUDA_HPP */

include/opencv2/core/cuda.inl.hpp

@@ -0,0 +1,631 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                          License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef OPENCV_CORE_CUDAINL_HPP
+#define OPENCV_CORE_CUDAINL_HPP
+
+#include "opencv2/core/cuda.hpp"
+
+//! @cond IGNORED
+
+namespace cv { namespace cuda {
+
+//===================================================================================
+// GpuMat
+//===================================================================================
+
+inline
+GpuMat::GpuMat(Allocator* allocator_)
+    : flags(0), rows(0), cols(0), step(0), data(0), refcount(0), datastart(0), dataend(0), allocator(allocator_)
+{}
+
+inline
+GpuMat::GpuMat(int rows_, int cols_, int type_, Allocator* allocator_)
+    : flags(0), rows(0), cols(0), step(0), data(0), refcount(0), datastart(0), dataend(0), allocator(allocator_)
+{
+    if (rows_ > 0 && cols_ > 0)
+        create(rows_, cols_, type_);
+}
+
+inline
+GpuMat::GpuMat(Size size_, int type_, Allocator* allocator_)
+    : flags(0), rows(0), cols(0), step(0), data(0), refcount(0), datastart(0), dataend(0), allocator(allocator_)
+{
+    if (size_.height > 0 && size_.width > 0)
+        create(size_.height, size_.width, type_);
+}
+
+inline
+GpuMat::GpuMat(int rows_, int cols_, int type_, Scalar s_, Allocator* allocator_)
+    : flags(0), rows(0), cols(0), step(0), data(0), refcount(0), datastart(0), dataend(0), allocator(allocator_)
+{
+    if (rows_ > 0 && cols_ > 0)
+    {
+        create(rows_, cols_, type_);
+        setTo(s_);
+    }
+}
+
+inline
+GpuMat::GpuMat(Size size_, int type_, Scalar s_, Allocator* allocator_)
+    : flags(0), rows(0), cols(0), step(0), data(0), refcount(0), datastart(0), dataend(0), allocator(allocator_)
+{
+    if (size_.height > 0 && size_.width > 0)
+    {
+        create(size_.height, size_.width, type_);
+        setTo(s_);
+    }
+}
+
+inline
+GpuMat::GpuMat(const GpuMat& m)
+    : flags(m.flags), rows(m.rows), cols(m.cols), step(m.step), data(m.data), refcount(m.refcount), datastart(m.datastart), dataend(m.dataend), allocator(m.allocator)
+{
+    if (refcount)
+        CV_XADD(refcount, 1);
+}
+
+inline
+GpuMat::GpuMat(InputArray arr, Allocator* allocator_) :
+    flags(0), rows(0), cols(0), step(0), data(0), refcount(0), datastart(0), dataend(0), allocator(allocator_)
+{
+    upload(arr);
+}
+
+inline
+GpuMat::~GpuMat()
+{
+    release();
+}
+
+inline
+GpuMat& GpuMat::operator =(const GpuMat& m)
+{
+    if (this != &m)
+    {
+        GpuMat temp(m);
+        swap(temp);
+    }
+
+    return *this;
+}
+
+inline
+void GpuMat::create(Size size_, int type_)
+{
+    create(size_.height, size_.width, type_);
+}
+
+inline
+void GpuMat::swap(GpuMat& b)
+{
+    std::swap(flags, b.flags);
+    std::swap(rows, b.rows);
+    std::swap(cols, b.cols);
+    std::swap(step, b.step);
+    std::swap(data, b.data);
+    std::swap(datastart, b.datastart);
+    std::swap(dataend, b.dataend);
+    std::swap(refcount, b.refcount);
+    std::swap(allocator, b.allocator);
+}
+
+inline
+GpuMat GpuMat::clone() const
+{
+    GpuMat m;
+    copyTo(m);
+    return m;
+}
+
+inline
+void GpuMat::copyTo(OutputArray dst, InputArray mask) const
+{
+    copyTo(dst, mask, Stream::Null());
+}
+
+inline
+GpuMat& GpuMat::setTo(Scalar s)
+{
+    return setTo(s, Stream::Null());
+}
+
+inline
+GpuMat& GpuMat::setTo(Scalar s, InputArray mask)
+{
+    return setTo(s, mask, Stream::Null());
+}
+
+inline
+void GpuMat::convertTo(OutputArray dst, int rtype) const
+{
+    convertTo(dst, rtype, Stream::Null());
+}
+
+inline
+void GpuMat::convertTo(OutputArray dst, int rtype, double alpha, double beta) const
+{
+    convertTo(dst, rtype, alpha, beta, Stream::Null());
+}
+
+inline
+void GpuMat::convertTo(OutputArray dst, int rtype, double alpha, Stream& stream) const
+{
+    convertTo(dst, rtype, alpha, 0.0, stream);
+}
+
+inline
+void GpuMat::assignTo(GpuMat& m, int _type) const
+{
+    if (_type < 0)
+        m = *this;
+    else
+        convertTo(m, _type);
+}
+
+inline
+uchar* GpuMat::ptr(int y)
+{
+    CV_DbgAssert( (unsigned)y < (unsigned)rows );
+    return data + step * y;
+}
+
+inline
+const uchar* GpuMat::ptr(int y) const
+{
+    CV_DbgAssert( (unsigned)y < (unsigned)rows );
+    return data + step * y;
+}
+
+template<typename _Tp> inline
+_Tp* GpuMat::ptr(int y)
+{
+    return (_Tp*)ptr(y);
+}
+
+template<typename _Tp> inline
+const _Tp* GpuMat::ptr(int y) const
+{
+    return (const _Tp*)ptr(y);
+}
+
+template <class T> inline
+GpuMat::operator PtrStepSz<T>() const
+{
+    return PtrStepSz<T>(rows, cols, (T*)data, step);
+}
+
+template <class T> inline
+GpuMat::operator PtrStep<T>() const
+{
+    return PtrStep<T>((T*)data, step);
+}
+
+inline
+GpuMat GpuMat::row(int y) const
+{
+    return GpuMat(*this, Range(y, y+1), Range::all());
+}
+
+inline
+GpuMat GpuMat::col(int x) const
+{
+    return GpuMat(*this, Range::all(), Range(x, x+1));
+}
+
+inline
+GpuMat GpuMat::rowRange(int startrow, int endrow) const
+{
+    return GpuMat(*this, Range(startrow, endrow), Range::all());
+}
+
+inline
+GpuMat GpuMat::rowRange(Range r) const
+{
+    return GpuMat(*this, r, Range::all());
+}
+
+inline
+GpuMat GpuMat::colRange(int startcol, int endcol) const
+{
+    return GpuMat(*this, Range::all(), Range(startcol, endcol));
+}
+
+inline
+GpuMat GpuMat::colRange(Range r) const
+{
+    return GpuMat(*this, Range::all(), r);
+}
+
+inline
+GpuMat GpuMat::operator ()(Range rowRange_, Range colRange_) const
+{
+    return GpuMat(*this, rowRange_, colRange_);
+}
+
+inline
+GpuMat GpuMat::operator ()(Rect roi) const
+{
+    return GpuMat(*this, roi);
+}
+
+inline
+bool GpuMat::isContinuous() const
+{
+    return (flags & Mat::CONTINUOUS_FLAG) != 0;
+}
+
+inline
+size_t GpuMat::elemSize() const
+{
+    return CV_ELEM_SIZE(flags);
+}
+
+inline
+size_t GpuMat::elemSize1() const
+{
+    return CV_ELEM_SIZE1(flags);
+}
+
+inline
+int GpuMat::type() const
+{
+    return CV_MAT_TYPE(flags);
+}
+
+inline
+int GpuMat::depth() const
+{
+    return CV_MAT_DEPTH(flags);
+}
+
+inline
+int GpuMat::channels() const
+{
+    return CV_MAT_CN(flags);
+}
+
+inline
+size_t GpuMat::step1() const
+{
+    return step / elemSize1();
+}
+
+inline
+Size GpuMat::size() const
+{
+    return Size(cols, rows);
+}
+
+inline
+bool GpuMat::empty() const
+{
+    return data == 0;
+}
+
+static inline
+GpuMat createContinuous(int rows, int cols, int type)
+{
+    GpuMat m;
+    createContinuous(rows, cols, type, m);
+    return m;
+}
+
+static inline
+void createContinuous(Size size, int type, OutputArray arr)
+{
+    createContinuous(size.height, size.width, type, arr);
+}
+
+static inline
+GpuMat createContinuous(Size size, int type)
+{
+    GpuMat m;
+    createContinuous(size, type, m);
+    return m;
+}
+
+static inline
+void ensureSizeIsEnough(Size size, int type, OutputArray arr)
+{
+    ensureSizeIsEnough(size.height, size.width, type, arr);
+}
+
+static inline
+void swap(GpuMat& a, GpuMat& b)
+{
+    a.swap(b);
+}
+
+//===================================================================================
+// HostMem
+//===================================================================================
+
+inline
+HostMem::HostMem(AllocType alloc_type_)
+    : flags(0), rows(0), cols(0), step(0), data(0), refcount(0), datastart(0), dataend(0), alloc_type(alloc_type_)
+{
+}
+
+inline
+HostMem::HostMem(const HostMem& m)
+    : flags(m.flags), rows(m.rows), cols(m.cols), step(m.step), data(m.data), refcount(m.refcount), datastart(m.datastart), dataend(m.dataend), alloc_type(m.alloc_type)
+{
+    if( refcount )
+        CV_XADD(refcount, 1);
+}
+
+inline
+HostMem::HostMem(int rows_, int cols_, int type_, AllocType alloc_type_)
+    : flags(0), rows(0), cols(0), step(0), data(0), refcount(0), datastart(0), dataend(0), alloc_type(alloc_type_)
+{
+    if (rows_ > 0 && cols_ > 0)
+        create(rows_, cols_, type_);
+}
+
+inline
+HostMem::HostMem(Size size_, int type_, AllocType alloc_type_)
+    : flags(0), rows(0), cols(0), step(0), data(0), refcount(0), datastart(0), dataend(0), alloc_type(alloc_type_)
+{
+    if (size_.height > 0 && size_.width > 0)
+        create(size_.height, size_.width, type_);
+}
+
+inline
+HostMem::HostMem(InputArray arr, AllocType alloc_type_)
+    : flags(0), rows(0), cols(0), step(0), data(0), refcount(0), datastart(0), dataend(0), alloc_type(alloc_type_)
+{
+    arr.getMat().copyTo(*this);
+}
+
+inline
+HostMem::~HostMem()
+{
+    release();
+}
+
+inline
+HostMem& HostMem::operator =(const HostMem& m)
+{
+    if (this != &m)
+    {
+        HostMem temp(m);
+        swap(temp);
+    }
+
+    return *this;
+}
+
+inline
+void HostMem::swap(HostMem& b)
+{
+    std::swap(flags, b.flags);
+    std::swap(rows, b.rows);
+    std::swap(cols, b.cols);
+    std::swap(step, b.step);
+    std::swap(data, b.data);
+    std::swap(datastart, b.datastart);
+    std::swap(dataend, b.dataend);
+    std::swap(refcount, b.refcount);
+    std::swap(alloc_type, b.alloc_type);
+}
+
+inline
+HostMem HostMem::clone() const
+{
+    HostMem m(size(), type(), alloc_type);
+    createMatHeader().copyTo(m);
+    return m;
+}
+
+inline
+void HostMem::create(Size size_, int type_)
+{
+    create(size_.height, size_.width, type_);
+}
+
+inline
+Mat HostMem::createMatHeader() const
+{
+    return Mat(size(), type(), data, step);
+}
+
+inline
+bool HostMem::isContinuous() const
+{
+    return (flags & Mat::CONTINUOUS_FLAG) != 0;
+}
+
+inline
+size_t HostMem::elemSize() const
+{
+    return CV_ELEM_SIZE(flags);
+}
+
+inline
+size_t HostMem::elemSize1() const
+{
+    return CV_ELEM_SIZE1(flags);
+}
+
+inline
+int HostMem::type() const
+{
+    return CV_MAT_TYPE(flags);
+}
+
+inline
+int HostMem::depth() const
+{
+    return CV_MAT_DEPTH(flags);
+}
+
+inline
+int HostMem::channels() const
+{
+    return CV_MAT_CN(flags);
+}
+
+inline
+size_t HostMem::step1() const
+{
+    return step / elemSize1();
+}
+
+inline
+Size HostMem::size() const
+{
+    return Size(cols, rows);
+}
+
+inline
+bool HostMem::empty() const
+{
+    return data == 0;
+}
+
+static inline
+void swap(HostMem& a, HostMem& b)
+{
+    a.swap(b);
+}
+
+//===================================================================================
+// Stream
+//===================================================================================
+
+inline
+Stream::Stream(const Ptr<Impl>& impl)
+    : impl_(impl)
+{
+}
+
+//===================================================================================
+// Event
+//===================================================================================
+
+inline
+Event::Event(const Ptr<Impl>& impl)
+    : impl_(impl)
+{
+}
+
+//===================================================================================
+// Initialization & Info
+//===================================================================================
+
+inline
+bool TargetArchs::has(int major, int minor)
+{
+    return hasPtx(major, minor) || hasBin(major, minor);
+}
+
+inline
+bool TargetArchs::hasEqualOrGreater(int major, int minor)
+{
+    return hasEqualOrGreaterPtx(major, minor) || hasEqualOrGreaterBin(major, minor);
+}
+
+inline
+DeviceInfo::DeviceInfo()
+{
+    device_id_ = getDevice();
+}
+
+inline
+DeviceInfo::DeviceInfo(int device_id)
+{
+    CV_Assert( device_id >= 0 && device_id < getCudaEnabledDeviceCount() );
+    device_id_ = device_id;
+}
+
+inline
+int DeviceInfo::deviceID() const
+{
+    return device_id_;
+}
+
+inline
+size_t DeviceInfo::freeMemory() const
+{
+    size_t _totalMemory = 0, _freeMemory = 0;
+    queryMemory(_totalMemory, _freeMemory);
+    return _freeMemory;
+}
+
+inline
+size_t DeviceInfo::totalMemory() const
+{
+    size_t _totalMemory = 0, _freeMemory = 0;
+    queryMemory(_totalMemory, _freeMemory);
+    return _totalMemory;
+}
+
+inline
+bool DeviceInfo::supports(FeatureSet feature_set) const
+{
+    int version = majorVersion() * 10 + minorVersion();
+    return version >= feature_set;
+}
+
+
+}} // namespace cv { namespace cuda {
+
+//===================================================================================
+// Mat
+//===================================================================================
+
+namespace cv {
+
+inline
+Mat::Mat(const cuda::GpuMat& m)
+    : flags(0), dims(0), rows(0), cols(0), data(0), datastart(0), dataend(0), datalimit(0), allocator(0), u(0), size(&rows)
+{
+    m.download(*this);
+}
+
+}
+
+//! @endcond
+
+#endif // OPENCV_CORE_CUDAINL_HPP

include/opencv2/core/cuda_devptrs.hpp

@@ -1,199 +0,0 @@
-/*M///////////////////////////////////////////////////////////////////////////////////////
-//
-//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
-//
-//  By downloading, copying, installing or using the software you agree to this license.
-//  If you do not agree to this license, do not download, install,
-//  copy or use the software.
-//
-//
-//                           License Agreement
-//                For Open Source Computer Vision Library
-//
-// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
-// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
-// Third party copyrights are property of their respective owners.
-//
-// Redistribution and use in source and binary forms, with or without modification,
-// are permitted provided that the following conditions are met:
-//
-//   * Redistribution's of source code must retain the above copyright notice,
-//     this list of conditions and the following disclaimer.
-//
-//   * Redistribution's in binary form must reproduce the above copyright notice,
-//     this list of conditions and the following disclaimer in the documentation
-//     and/or other materials provided with the distribution.
-//
-//   * The name of the copyright holders may not be used to endorse or promote products
-//     derived from this software without specific prior written permission.
-//
-// This software is provided by the copyright holders and contributors "as is" and
-// any express or implied warranties, including, but not limited to, the implied
-// warranties of merchantability and fitness for a particular purpose are disclaimed.
-// In no event shall the Intel Corporation or contributors be liable for any direct,
-// indirect, incidental, special, exemplary, or consequential damages
-// (including, but not limited to, procurement of substitute goods or services;
-// loss of use, data, or profits; or business interruption) however caused
-// and on any theory of liability, whether in contract, strict liability,
-// or tort (including negligence or otherwise) arising in any way out of
-// the use of this software, even if advised of the possibility of such damage.
-//
-//M*/
-
-#ifndef __OPENCV_CORE_DEVPTRS_HPP__
-#define __OPENCV_CORE_DEVPTRS_HPP__
-
-#ifdef __cplusplus
-
-#ifdef __CUDACC__
-    #define __CV_GPU_HOST_DEVICE__ __host__ __device__ __forceinline__
-#else
-    #define __CV_GPU_HOST_DEVICE__
-#endif
-
-namespace cv
-{
-    namespace gpu
-    {
-        // Simple lightweight structures that encapsulates information about an image on device.
-        // It is intended to pass to nvcc-compiled code. GpuMat depends on headers that nvcc can't compile
-
-        template <bool expr> struct StaticAssert;
-        template <> struct StaticAssert<true> {static __CV_GPU_HOST_DEVICE__ void check(){}};
-
-        template<typename T> struct DevPtr
-        {
-            typedef T elem_type;
-            typedef int index_type;
-
-            enum { elem_size = sizeof(elem_type) };
-
-            T* data;
-
-            __CV_GPU_HOST_DEVICE__ DevPtr() : data(0) {}
-            __CV_GPU_HOST_DEVICE__ DevPtr(T* data_) : data(data_) {}
-
-            __CV_GPU_HOST_DEVICE__ size_t elemSize() const { return elem_size; }
-            __CV_GPU_HOST_DEVICE__ operator       T*()       { return data; }
-            __CV_GPU_HOST_DEVICE__ operator const T*() const { return data; }
-        };
-
-        template<typename T> struct PtrSz : public DevPtr<T>
-        {
-            __CV_GPU_HOST_DEVICE__ PtrSz() : size(0) {}
-            __CV_GPU_HOST_DEVICE__ PtrSz(T* data_, size_t size_) : DevPtr<T>(data_), size(size_) {}
-
-            size_t size;
-        };
-
-        template<typename T> struct PtrStep : public DevPtr<T>
-        {
-            __CV_GPU_HOST_DEVICE__ PtrStep() : step(0) {}
-            __CV_GPU_HOST_DEVICE__ PtrStep(T* data_, size_t step_) : DevPtr<T>(data_), step(step_) {}
-
-            /** \brief stride between two consecutive rows in bytes. Step is stored always and everywhere in bytes!!! */
-            size_t step;
-
-            __CV_GPU_HOST_DEVICE__       T* ptr(int y = 0)       { return (      T*)( (      char*)DevPtr<T>::data + y * step); }
-            __CV_GPU_HOST_DEVICE__ const T* ptr(int y = 0) const { return (const T*)( (const char*)DevPtr<T>::data + y * step); }
-
-            __CV_GPU_HOST_DEVICE__       T& operator ()(int y, int x)       { return ptr(y)[x]; }
-            __CV_GPU_HOST_DEVICE__ const T& operator ()(int y, int x) const { return ptr(y)[x]; }
-        };
-
-        template <typename T> struct PtrStepSz : public PtrStep<T>
-        {
-            __CV_GPU_HOST_DEVICE__ PtrStepSz() : cols(0), rows(0) {}
-            __CV_GPU_HOST_DEVICE__ PtrStepSz(int rows_, int cols_, T* data_, size_t step_)
-                : PtrStep<T>(data_, step_), cols(cols_), rows(rows_) {}
-
-            template <typename U>
-            explicit PtrStepSz(const PtrStepSz<U>& d) : PtrStep<T>((T*)d.data, d.step), cols(d.cols), rows(d.rows){}
-
-            int cols;
-            int rows;
-        };
-
-        typedef PtrStepSz<unsigned char> PtrStepSzb;
-        typedef PtrStepSz<float> PtrStepSzf;
-        typedef PtrStepSz<int> PtrStepSzi;
-
-        typedef PtrStep<unsigned char> PtrStepb;
-        typedef PtrStep<float> PtrStepf;
-        typedef PtrStep<int> PtrStepi;
-
-
-#if defined __GNUC__
-    #define __CV_GPU_DEPR_BEFORE__
-    #define __CV_GPU_DEPR_AFTER__ __attribute__ ((deprecated))
-#elif defined(__MSVC__) //|| defined(__CUDACC__)
-    #pragma deprecated(DevMem2D_)
-    #define __CV_GPU_DEPR_BEFORE__ __declspec(deprecated)
-    #define __CV_GPU_DEPR_AFTER__
-#else
-    #define __CV_GPU_DEPR_BEFORE__
-    #define __CV_GPU_DEPR_AFTER__
-#endif
-
-        template <typename T> struct __CV_GPU_DEPR_BEFORE__ DevMem2D_ : public PtrStepSz<T>
-        {
-            DevMem2D_() {}
-            DevMem2D_(int rows_, int cols_, T* data_, size_t step_) : PtrStepSz<T>(rows_, cols_, data_, step_) {}
-
-            template <typename U>
-            explicit __CV_GPU_DEPR_BEFORE__ DevMem2D_(const DevMem2D_<U>& d) : PtrStepSz<T>(d.rows, d.cols, (T*)d.data, d.step) {}
-        } __CV_GPU_DEPR_AFTER__ ;
-
-        typedef DevMem2D_<unsigned char> DevMem2Db;
-        typedef DevMem2Db DevMem2D;
-        typedef DevMem2D_<float> DevMem2Df;
-        typedef DevMem2D_<int> DevMem2Di;
-
-        template<typename T> struct PtrElemStep_ : public PtrStep<T>
-        {
-            PtrElemStep_(const DevMem2D_<T>& mem) : PtrStep<T>(mem.data, mem.step)
-            {
-                StaticAssert<256 % sizeof(T) == 0>::check();
-
-                PtrStep<T>::step /= PtrStep<T>::elem_size;
-            }
-            __CV_GPU_HOST_DEVICE__ T* ptr(int y = 0) { return PtrStep<T>::data + y * PtrStep<T>::step; }
-            __CV_GPU_HOST_DEVICE__ const T* ptr(int y = 0) const { return PtrStep<T>::data + y * PtrStep<T>::step; }
-
-            __CV_GPU_HOST_DEVICE__ T& operator ()(int y, int x) { return ptr(y)[x]; }
-            __CV_GPU_HOST_DEVICE__ const T& operator ()(int y, int x) const { return ptr(y)[x]; }
-        };
-
-        template<typename T> struct PtrStep_ : public PtrStep<T>
-        {
-            PtrStep_() {}
-            PtrStep_(const DevMem2D_<T>& mem) : PtrStep<T>(mem.data, mem.step) {}
-        };
-
-        typedef PtrElemStep_<unsigned char> PtrElemStep;
-        typedef PtrElemStep_<float> PtrElemStepf;
-        typedef PtrElemStep_<int> PtrElemStepi;
-
-//#undef __CV_GPU_DEPR_BEFORE__
-//#undef __CV_GPU_DEPR_AFTER__
-
-        namespace device
-        {
-            using cv::gpu::PtrSz;
-            using cv::gpu::PtrStep;
-            using cv::gpu::PtrStepSz;
-
-            using cv::gpu::PtrStepSzb;
-            using cv::gpu::PtrStepSzf;
-            using cv::gpu::PtrStepSzi;
-
-            using cv::gpu::PtrStepb;
-            using cv::gpu::PtrStepf;
-            using cv::gpu::PtrStepi;
-        }
-    }
-}
-
-#endif // __cplusplus
-
-#endif /* __OPENCV_CORE_DEVPTRS_HPP__ */

include/opencv2/core/cuda_stream_accessor.hpp

@@ -0,0 +1,86 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                           License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef OPENCV_CORE_CUDA_STREAM_ACCESSOR_HPP
+#define OPENCV_CORE_CUDA_STREAM_ACCESSOR_HPP
+
+#ifndef __cplusplus
+#  error cuda_stream_accessor.hpp header must be compiled as C++
+#endif
+
+/** @file cuda_stream_accessor.hpp
+ * This is only header file that depends on CUDA Runtime API. All other headers are independent.
+ */
+
+#include <cuda_runtime.h>
+#include "opencv2/core/cuda.hpp"
+
+namespace cv
+{
+    namespace cuda
+    {
+
+//! @addtogroup cudacore_struct
+//! @{
+
+        /** @brief Class that enables getting cudaStream_t from cuda::Stream
+         */
+        struct StreamAccessor
+        {
+            CV_EXPORTS static cudaStream_t getStream(const Stream& stream);
+            CV_EXPORTS static Stream wrapStream(cudaStream_t stream);
+        };
+
+        /** @brief Class that enables getting cudaEvent_t from cuda::Event
+         */
+        struct EventAccessor
+        {
+            CV_EXPORTS static cudaEvent_t getEvent(const Event& event);
+            CV_EXPORTS static Event wrapEvent(cudaEvent_t event);
+        };
+
+//! @}
+
+    }
+}
+
+#endif /* OPENCV_CORE_CUDA_STREAM_ACCESSOR_HPP */

include/opencv2/core/cuda_types.hpp

@@ -0,0 +1,135 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                           License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef OPENCV_CORE_CUDA_TYPES_HPP
+#define OPENCV_CORE_CUDA_TYPES_HPP
+
+#ifndef __cplusplus
+#  error cuda_types.hpp header must be compiled as C++
+#endif
+
+/** @file
+ * @deprecated Use @ref cudev instead.
+ */
+
+//! @cond IGNORED
+
+#ifdef __CUDACC__
+    #define __CV_CUDA_HOST_DEVICE__ __host__ __device__ __forceinline__
+#else
+    #define __CV_CUDA_HOST_DEVICE__
+#endif
+
+namespace cv
+{
+    namespace cuda
+    {
+
+        // Simple lightweight structures that encapsulates information about an image on device.
+        // It is intended to pass to nvcc-compiled code. GpuMat depends on headers that nvcc can't compile
+
+        template <typename T> struct DevPtr
+        {
+            typedef T elem_type;
+            typedef int index_type;
+
+            enum { elem_size = sizeof(elem_type) };
+
+            T* data;
+
+            __CV_CUDA_HOST_DEVICE__ DevPtr() : data(0) {}
+            __CV_CUDA_HOST_DEVICE__ DevPtr(T* data_) : data(data_) {}
+
+            __CV_CUDA_HOST_DEVICE__ size_t elemSize() const { return elem_size; }
+            __CV_CUDA_HOST_DEVICE__ operator       T*()       { return data; }
+            __CV_CUDA_HOST_DEVICE__ operator const T*() const { return data; }
+        };
+
+        template <typename T> struct PtrSz : public DevPtr<T>
+        {
+            __CV_CUDA_HOST_DEVICE__ PtrSz() : size(0) {}
+            __CV_CUDA_HOST_DEVICE__ PtrSz(T* data_, size_t size_) : DevPtr<T>(data_), size(size_) {}
+
+            size_t size;
+        };
+
+        template <typename T> struct PtrStep : public DevPtr<T>
+        {
+            __CV_CUDA_HOST_DEVICE__ PtrStep() : step(0) {}
+            __CV_CUDA_HOST_DEVICE__ PtrStep(T* data_, size_t step_) : DevPtr<T>(data_), step(step_) {}
+
+            size_t step;
+
+            __CV_CUDA_HOST_DEVICE__       T* ptr(int y = 0)       { return (      T*)( (      char*)DevPtr<T>::data + y * step); }
+            __CV_CUDA_HOST_DEVICE__ const T* ptr(int y = 0) const { return (const T*)( (const char*)DevPtr<T>::data + y * step); }
+
+            __CV_CUDA_HOST_DEVICE__       T& operator ()(int y, int x)       { return ptr(y)[x]; }
+            __CV_CUDA_HOST_DEVICE__ const T& operator ()(int y, int x) const { return ptr(y)[x]; }
+        };
+
+        template <typename T> struct PtrStepSz : public PtrStep<T>
+        {
+            __CV_CUDA_HOST_DEVICE__ PtrStepSz() : cols(0), rows(0) {}
+            __CV_CUDA_HOST_DEVICE__ PtrStepSz(int rows_, int cols_, T* data_, size_t step_)
+                : PtrStep<T>(data_, step_), cols(cols_), rows(rows_) {}
+
+            template <typename U>
+            explicit PtrStepSz(const PtrStepSz<U>& d) : PtrStep<T>((T*)d.data, d.step), cols(d.cols), rows(d.rows){}
+
+            int cols;
+            int rows;
+        };
+
+        typedef PtrStepSz<unsigned char> PtrStepSzb;
+        typedef PtrStepSz<float> PtrStepSzf;
+        typedef PtrStepSz<int> PtrStepSzi;
+
+        typedef PtrStep<unsigned char> PtrStepb;
+        typedef PtrStep<float> PtrStepf;
+        typedef PtrStep<int> PtrStepi;
+
+    }
+}
+
+//! @endcond
+
+#endif /* OPENCV_CORE_CUDA_TYPES_HPP */

include/opencv2/core/cvdef.h

@@ -0,0 +1,481 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                          License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
+// Copyright (C) 2015, Itseez Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef OPENCV_CORE_CVDEF_H
+#define OPENCV_CORE_CVDEF_H
+
+//! @addtogroup core_utils
+//! @{
+
+#if !defined _CRT_SECURE_NO_DEPRECATE && defined _MSC_VER && _MSC_VER > 1300
+#  define _CRT_SECURE_NO_DEPRECATE /* to avoid multiple Visual Studio warnings */
+#endif
+
+// undef problematic defines sometimes defined by system headers (windows.h in particular)
+#undef small
+#undef min
+#undef max
+#undef abs
+#undef Complex
+
+#if !defined _CRT_SECURE_NO_DEPRECATE && defined _MSC_VER && _MSC_VER > 1300
+#  define _CRT_SECURE_NO_DEPRECATE /* to avoid multiple Visual Studio warnings */
+#endif
+
+#include <limits.h>
+#include "opencv2/core/hal/interface.h"
+
+#if defined __ICL
+#  define CV_ICC   __ICL
+#elif defined __ICC
+#  define CV_ICC   __ICC
+#elif defined __ECL
+#  define CV_ICC   __ECL
+#elif defined __ECC
+#  define CV_ICC   __ECC
+#elif defined __INTEL_COMPILER
+#  define CV_ICC   __INTEL_COMPILER
+#endif
+
+#ifndef CV_INLINE
+#  if defined __cplusplus
+#    define CV_INLINE static inline
+#  elif defined _MSC_VER
+#    define CV_INLINE __inline
+#  else
+#    define CV_INLINE static
+#  endif
+#endif
+
+#if defined CV_ICC && !defined CV_ENABLE_UNROLLED
+#  define CV_ENABLE_UNROLLED 0
+#else
+#  define CV_ENABLE_UNROLLED 1
+#endif
+
+#ifdef __GNUC__
+#  define CV_DECL_ALIGNED(x) __attribute__ ((aligned (x)))
+#elif defined _MSC_VER
+#  define CV_DECL_ALIGNED(x) __declspec(align(x))
+#else
+#  define CV_DECL_ALIGNED(x)
+#endif
+
+/* CPU features and intrinsics support */
+#define CV_CPU_NONE             0
+#define CV_CPU_MMX              1
+#define CV_CPU_SSE              2
+#define CV_CPU_SSE2             3
+#define CV_CPU_SSE3             4
+#define CV_CPU_SSSE3            5
+#define CV_CPU_SSE4_1           6
+#define CV_CPU_SSE4_2           7
+#define CV_CPU_POPCNT           8
+#define CV_CPU_FP16             9
+#define CV_CPU_AVX              10
+#define CV_CPU_AVX2             11
+#define CV_CPU_FMA3             12
+
+#define CV_CPU_AVX_512F         13
+#define CV_CPU_AVX_512BW        14
+#define CV_CPU_AVX_512CD        15
+#define CV_CPU_AVX_512DQ        16
+#define CV_CPU_AVX_512ER        17
+#define CV_CPU_AVX_512IFMA512   18
+#define CV_CPU_AVX_512PF        19
+#define CV_CPU_AVX_512VBMI      20
+#define CV_CPU_AVX_512VL        21
+
+#define CV_CPU_NEON   100
+
+// when adding to this list remember to update the following enum
+#define CV_HARDWARE_MAX_FEATURE 255
+
+/** @brief Available CPU features.
+*/
+enum CpuFeatures {
+    CPU_MMX             = 1,
+    CPU_SSE             = 2,
+    CPU_SSE2            = 3,
+    CPU_SSE3            = 4,
+    CPU_SSSE3           = 5,
+    CPU_SSE4_1          = 6,
+    CPU_SSE4_2          = 7,
+    CPU_POPCNT          = 8,
+    CPU_FP16            = 9,
+    CPU_AVX             = 10,
+    CPU_AVX2            = 11,
+    CPU_FMA3            = 12,
+
+    CPU_AVX_512F        = 13,
+    CPU_AVX_512BW       = 14,
+    CPU_AVX_512CD       = 15,
+    CPU_AVX_512DQ       = 16,
+    CPU_AVX_512ER       = 17,
+    CPU_AVX_512IFMA512  = 18,
+    CPU_AVX_512PF       = 19,
+    CPU_AVX_512VBMI     = 20,
+    CPU_AVX_512VL       = 21,
+
+    CPU_NEON            = 100
+};
+
+// do not include SSE/AVX/NEON headers for NVCC compiler
+#ifndef __CUDACC__
+
+#if defined __SSE2__ || defined _M_X64 || (defined _M_IX86_FP && _M_IX86_FP >= 2)
+#  include <emmintrin.h>
+#  define CV_MMX 1
+#  define CV_SSE 1
+#  define CV_SSE2 1
+#  if defined __SSE3__ || (defined _MSC_VER && _MSC_VER >= 1500)
+#    include <pmmintrin.h>
+#    define CV_SSE3 1
+#  endif
+#  if defined __SSSE3__  || (defined _MSC_VER && _MSC_VER >= 1500)
+#    include <tmmintrin.h>
+#    define CV_SSSE3 1
+#  endif
+#  if defined __SSE4_1__ || (defined _MSC_VER && _MSC_VER >= 1500)
+#    include <smmintrin.h>
+#    define CV_SSE4_1 1
+#  endif
+#  if defined __SSE4_2__ || (defined _MSC_VER && _MSC_VER >= 1500)
+#    include <nmmintrin.h>
+#    define CV_SSE4_2 1
+#  endif
+#  if defined __POPCNT__ || (defined _MSC_VER && _MSC_VER >= 1500)
+#    ifdef _MSC_VER
+#      include <nmmintrin.h>
+#    else
+#      include <popcntintrin.h>
+#    endif
+#    define CV_POPCNT 1
+#  endif
+#  if defined __AVX__ || (defined _MSC_VER && _MSC_VER >= 1600 && 0)
+// MS Visual Studio 2010 (2012?) has no macro pre-defined to identify the use of /arch:AVX
+// See: http://connect.microsoft.com/VisualStudio/feedback/details/605858/arch-avx-should-define-a-predefined-macro-in-x64-and-set-a-unique-value-for-m-ix86-fp-in-win32
+#    include <immintrin.h>
+#    define CV_AVX 1
+#    if defined(_XCR_XFEATURE_ENABLED_MASK)
+#      define __xgetbv() _xgetbv(_XCR_XFEATURE_ENABLED_MASK)
+#    else
+#      define __xgetbv() 0
+#    endif
+#  endif
+#  if defined __AVX2__ || (defined _MSC_VER && _MSC_VER >= 1800 && 0)
+#    include <immintrin.h>
+#    define CV_AVX2 1
+#    if defined __FMA__
+#      define CV_FMA3 1
+#    endif
+#  endif
+#endif
+
+#if (defined WIN32 || defined _WIN32) && defined(_M_ARM)
+# include <Intrin.h>
+# include <arm_neon.h>
+# define CV_NEON 1
+# define CPU_HAS_NEON_FEATURE (true)
+#elif defined(__ARM_NEON__) || (defined (__ARM_NEON) && defined(__aarch64__))
+#  include <arm_neon.h>
+#  define CV_NEON 1
+#endif
+
+#if defined __GNUC__ && defined __arm__ && (defined __ARM_PCS_VFP || defined __ARM_VFPV3__ || defined __ARM_NEON__) && !defined __SOFTFP__
+#  define CV_VFP 1
+#endif
+
+#endif // __CUDACC__
+
+#ifndef CV_POPCNT
+#define CV_POPCNT 0
+#endif
+#ifndef CV_MMX
+#  define CV_MMX 0
+#endif
+#ifndef CV_SSE
+#  define CV_SSE 0
+#endif
+#ifndef CV_SSE2
+#  define CV_SSE2 0
+#endif
+#ifndef CV_SSE3
+#  define CV_SSE3 0
+#endif
+#ifndef CV_SSSE3
+#  define CV_SSSE3 0
+#endif
+#ifndef CV_SSE4_1
+#  define CV_SSE4_1 0
+#endif
+#ifndef CV_SSE4_2
+#  define CV_SSE4_2 0
+#endif
+#ifndef CV_AVX
+#  define CV_AVX 0
+#endif
+#ifndef CV_AVX2
+#  define CV_AVX2 0
+#endif
+#ifndef CV_FMA3
+#  define CV_FMA3 0
+#endif
+#ifndef CV_AVX_512F
+#  define CV_AVX_512F 0
+#endif
+#ifndef CV_AVX_512BW
+#  define CV_AVX_512BW 0
+#endif
+#ifndef CV_AVX_512CD
+#  define CV_AVX_512CD 0
+#endif
+#ifndef CV_AVX_512DQ
+#  define CV_AVX_512DQ 0
+#endif
+#ifndef CV_AVX_512ER
+#  define CV_AVX_512ER 0
+#endif
+#ifndef CV_AVX_512IFMA512
+#  define CV_AVX_512IFMA512 0
+#endif
+#ifndef CV_AVX_512PF
+#  define CV_AVX_512PF 0
+#endif
+#ifndef CV_AVX_512VBMI
+#  define CV_AVX_512VBMI 0
+#endif
+#ifndef CV_AVX_512VL
+#  define CV_AVX_512VL 0
+#endif
+
+#ifndef CV_NEON
+#  define CV_NEON 0
+#endif
+
+#ifndef CV_VFP
+#  define CV_VFP 0
+#endif
+
+/* fundamental constants */
+#define CV_PI   3.1415926535897932384626433832795
+#define CV_2PI 6.283185307179586476925286766559
+#define CV_LOG2 0.69314718055994530941723212145818
+
+#if defined __ARM_FP16_FORMAT_IEEE \
+    && !defined __CUDACC__
+#  define CV_FP16_TYPE 1
+#else
+#  define CV_FP16_TYPE 0
+#endif
+
+typedef union Cv16suf
+{
+    short i;
+#if CV_FP16_TYPE
+    __fp16 h;
+#endif
+    struct _fp16Format
+    {
+        unsigned int significand : 10;
+        unsigned int exponent    : 5;
+        unsigned int sign        : 1;
+    } fmt;
+}
+Cv16suf;
+
+typedef union Cv32suf
+{
+    int i;
+    unsigned u;
+    float f;
+    struct _fp32Format
+    {
+        unsigned int significand : 23;
+        unsigned int exponent    : 8;
+        unsigned int sign        : 1;
+    } fmt;
+}
+Cv32suf;
+
+typedef union Cv64suf
+{
+    int64 i;
+    uint64 u;
+    double f;
+}
+Cv64suf;
+
+#define OPENCV_ABI_COMPATIBILITY 300
+
+#ifdef __OPENCV_BUILD
+#  define DISABLE_OPENCV_24_COMPATIBILITY
+#endif
+
+#if (defined WIN32 || defined _WIN32 || defined WINCE || defined __CYGWIN__) && defined CVAPI_EXPORTS
+#  define CV_EXPORTS __declspec(dllexport)
+#elif defined __GNUC__ && __GNUC__ >= 4
+#  define CV_EXPORTS __attribute__ ((visibility ("default")))
+#else
+#  define CV_EXPORTS
+#endif
+
+#ifndef CV_EXTERN_C
+#  ifdef __cplusplus
+#    define CV_EXTERN_C extern "C"
+#  else
+#    define CV_EXTERN_C
+#  endif
+#endif
+
+/* special informative macros for wrapper generators */
+#define CV_EXPORTS_W CV_EXPORTS
+#define CV_EXPORTS_W_SIMPLE CV_EXPORTS
+#define CV_EXPORTS_AS(synonym) CV_EXPORTS
+#define CV_EXPORTS_W_MAP CV_EXPORTS
+#define CV_IN_OUT
+#define CV_OUT
+#define CV_PROP
+#define CV_PROP_RW
+#define CV_WRAP
+#define CV_WRAP_AS(synonym)
+
+/****************************************************************************************\
+*                                  Matrix type (Mat)                                     *
+\****************************************************************************************/
+
+#define CV_MAT_CN_MASK          ((CV_CN_MAX - 1) << CV_CN_SHIFT)
+#define CV_MAT_CN(flags)        ((((flags) & CV_MAT_CN_MASK) >> CV_CN_SHIFT) + 1)
+#define CV_MAT_TYPE_MASK        (CV_DEPTH_MAX*CV_CN_MAX - 1)
+#define CV_MAT_TYPE(flags)      ((flags) & CV_MAT_TYPE_MASK)
+#define CV_MAT_CONT_FLAG_SHIFT  14
+#define CV_MAT_CONT_FLAG        (1 << CV_MAT_CONT_FLAG_SHIFT)
+#define CV_IS_MAT_CONT(flags)   ((flags) & CV_MAT_CONT_FLAG)
+#define CV_IS_CONT_MAT          CV_IS_MAT_CONT
+#define CV_SUBMAT_FLAG_SHIFT    15
+#define CV_SUBMAT_FLAG          (1 << CV_SUBMAT_FLAG_SHIFT)
+#define CV_IS_SUBMAT(flags)     ((flags) & CV_MAT_SUBMAT_FLAG)
+
+/** Size of each channel item,
+   0x124489 = 1000 0100 0100 0010 0010 0001 0001 ~ array of sizeof(arr_type_elem) */
+#define CV_ELEM_SIZE1(type) \
+    ((((sizeof(size_t)<<28)|0x8442211) >> CV_MAT_DEPTH(type)*4) & 15)
+
+/** 0x3a50 = 11 10 10 01 01 00 00 ~ array of log2(sizeof(arr_type_elem)) */
+#define CV_ELEM_SIZE(type) \
+    (CV_MAT_CN(type) << ((((sizeof(size_t)/4+1)*16384|0x3a50) >> CV_MAT_DEPTH(type)*2) & 3))
+
+#ifndef MIN
+#  define MIN(a,b)  ((a) > (b) ? (b) : (a))
+#endif
+
+#ifndef MAX
+#  define MAX(a,b)  ((a) < (b) ? (b) : (a))
+#endif
+
+/****************************************************************************************\
+*          exchange-add operation for atomic operations on reference counters            *
+\****************************************************************************************/
+
+#ifdef CV_XADD
+  // allow to use user-defined macro
+#elif defined __GNUC__
+#  if defined __clang__ && __clang_major__ >= 3 && !defined __ANDROID__ && !defined __EMSCRIPTEN__ && !defined(__CUDACC__)
+#    ifdef __ATOMIC_ACQ_REL
+#      define CV_XADD(addr, delta) __c11_atomic_fetch_add((_Atomic(int)*)(addr), delta, __ATOMIC_ACQ_REL)
+#    else
+#      define CV_XADD(addr, delta) __atomic_fetch_add((_Atomic(int)*)(addr), delta, 4)
+#    endif
+#  else
+#    if defined __ATOMIC_ACQ_REL && !defined __clang__
+       // version for gcc >= 4.7
+#      define CV_XADD(addr, delta) (int)__atomic_fetch_add((unsigned*)(addr), (unsigned)(delta), __ATOMIC_ACQ_REL)
+#    else
+#      define CV_XADD(addr, delta) (int)__sync_fetch_and_add((unsigned*)(addr), (unsigned)(delta))
+#    endif
+#  endif
+#elif defined _MSC_VER && !defined RC_INVOKED
+#  include <intrin.h>
+#  define CV_XADD(addr, delta) (int)_InterlockedExchangeAdd((long volatile*)addr, delta)
+#else
+   CV_INLINE CV_XADD(int* addr, int delta) { int tmp = *addr; *addr += delta; return tmp; }
+#endif
+
+
+/****************************************************************************************\
+*                                  CV_NORETURN attribute                                 *
+\****************************************************************************************/
+
+#ifndef CV_NORETURN
+#  if defined(__GNUC__)
+#    define CV_NORETURN __attribute__((__noreturn__))
+#  elif defined(_MSC_VER) && (_MSC_VER >= 1300)
+#    define CV_NORETURN __declspec(noreturn)
+#  else
+#    define CV_NORETURN /* nothing by default */
+#  endif
+#endif
+
+
+/****************************************************************************************\
+*                                    C++ Move semantics                                  *
+\****************************************************************************************/
+
+#ifndef CV_CXX_MOVE_SEMANTICS
+#  if __cplusplus >= 201103L || defined(__GXX_EXPERIMENTAL_CXX0X__) || defined(_MSC_VER) && _MSC_VER >= 1600
+#    define CV_CXX_MOVE_SEMANTICS 1
+#  elif defined(__clang)
+#    if __has_feature(cxx_rvalue_references)
+#      define CV_CXX_MOVE_SEMANTICS 1
+#    endif
+#  endif
+#else
+#  if CV_CXX_MOVE_SEMANTICS == 0
+#    undef CV_CXX_MOVE_SEMANTICS
+#  endif
+#endif
+
+//! @}
+
+#endif // OPENCV_CORE_CVDEF_H

include/opencv2/core/cvstd.inl.hpp

@@ -0,0 +1,267 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                          License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef OPENCV_CORE_CVSTDINL_HPP
+#define OPENCV_CORE_CVSTDINL_HPP
+
+#ifndef OPENCV_NOSTL
+#  include <complex>
+#  include <ostream>
+#endif
+
+//! @cond IGNORED
+
+namespace cv
+{
+#ifndef OPENCV_NOSTL
+
+template<typename _Tp> class DataType< std::complex<_Tp> >
+{
+public:
+    typedef std::complex<_Tp>  value_type;
+    typedef value_type         work_type;
+    typedef _Tp                channel_type;
+
+    enum { generic_type = 0,
+           depth        = DataType<channel_type>::depth,
+           channels     = 2,
+           fmt          = DataType<channel_type>::fmt + ((channels - 1) << 8),
+           type         = CV_MAKETYPE(depth, channels) };
+
+    typedef Vec<channel_type, channels> vec_type;
+};
+
+inline
+String::String(const std::string& str)
+    : cstr_(0), len_(0)
+{
+    if (!str.empty())
+    {
+        size_t len = str.size();
+        memcpy(allocate(len), str.c_str(), len);
+    }
+}
+
+inline
+String::String(const std::string& str, size_t pos, size_t len)
+    : cstr_(0), len_(0)
+{
+    size_t strlen = str.size();
+    pos = min(pos, strlen);
+    len = min(strlen - pos, len);
+    if (!len) return;
+    memcpy(allocate(len), str.c_str() + pos, len);
+}
+
+inline
+String& String::operator = (const std::string& str)
+{
+    deallocate();
+    if (!str.empty())
+    {
+        size_t len = str.size();
+        memcpy(allocate(len), str.c_str(), len);
+    }
+    return *this;
+}
+
+inline
+String& String::operator += (const std::string& str)
+{
+    *this = *this + str;
+    return *this;
+}
+
+inline
+String::operator std::string() const
+{
+    return std::string(cstr_, len_);
+}
+
+inline
+String operator + (const String& lhs, const std::string& rhs)
+{
+    String s;
+    size_t rhslen = rhs.size();
+    s.allocate(lhs.len_ + rhslen);
+    memcpy(s.cstr_, lhs.cstr_, lhs.len_);
+    memcpy(s.cstr_ + lhs.len_, rhs.c_str(), rhslen);
+    return s;
+}
+
+inline
+String operator + (const std::string& lhs, const String& rhs)
+{
+    String s;
+    size_t lhslen = lhs.size();
+    s.allocate(lhslen + rhs.len_);
+    memcpy(s.cstr_, lhs.c_str(), lhslen);
+    memcpy(s.cstr_ + lhslen, rhs.cstr_, rhs.len_);
+    return s;
+}
+
+inline
+FileNode::operator std::string() const
+{
+    String value;
+    read(*this, value, value);
+    return value;
+}
+
+template<> inline
+void operator >> (const FileNode& n, std::string& value)
+{
+    String val;
+    read(n, val, val);
+    value = val;
+}
+
+template<> inline
+FileStorage& operator << (FileStorage& fs, const std::string& value)
+{
+    return fs << cv::String(value);
+}
+
+static inline
+std::ostream& operator << (std::ostream& os, const String& str)
+{
+    return os << str.c_str();
+}
+
+static inline
+std::ostream& operator << (std::ostream& out, Ptr<Formatted> fmtd)
+{
+    fmtd->reset();
+    for(const char* str = fmtd->next(); str; str = fmtd->next())
+        out << str;
+    return out;
+}
+
+static inline
+std::ostream& operator << (std::ostream& out, const Mat& mtx)
+{
+    return out << Formatter::get()->format(mtx);
+}
+
+template<typename _Tp> static inline
+std::ostream& operator << (std::ostream& out, const std::vector<Point_<_Tp> >& vec)
+{
+    return out << Formatter::get()->format(Mat(vec));
+}
+
+
+template<typename _Tp> static inline
+std::ostream& operator << (std::ostream& out, const std::vector<Point3_<_Tp> >& vec)
+{
+    return out << Formatter::get()->format(Mat(vec));
+}
+
+
+template<typename _Tp, int m, int n> static inline
+std::ostream& operator << (std::ostream& out, const Matx<_Tp, m, n>& matx)
+{
+    return out << Formatter::get()->format(Mat(matx));
+}
+
+template<typename _Tp> static inline
+std::ostream& operator << (std::ostream& out, const Point_<_Tp>& p)
+{
+    out << "[" << p.x << ", " << p.y << "]";
+    return out;
+}
+
+template<typename _Tp> static inline
+std::ostream& operator << (std::ostream& out, const Point3_<_Tp>& p)
+{
+    out << "[" << p.x << ", " << p.y << ", " << p.z << "]";
+    return out;
+}
+
+template<typename _Tp, int n> static inline
+std::ostream& operator << (std::ostream& out, const Vec<_Tp, n>& vec)
+{
+    out << "[";
+#ifdef _MSC_VER
+#pragma warning( push )
+#pragma warning( disable: 4127 )
+#endif
+    if(Vec<_Tp, n>::depth < CV_32F)
+#ifdef _MSC_VER
+#pragma warning( pop )
+#endif
+    {
+        for (int i = 0; i < n - 1; ++i) {
+            out << (int)vec[i] << ", ";
+        }
+        out << (int)vec[n-1] << "]";
+    }
+    else
+    {
+        for (int i = 0; i < n - 1; ++i) {
+            out << vec[i] << ", ";
+        }
+        out << vec[n-1] << "]";
+    }
+
+    return out;
+}
+
+template<typename _Tp> static inline
+std::ostream& operator << (std::ostream& out, const Size_<_Tp>& size)
+{
+    return out << "[" << size.width << " x " << size.height << "]";
+}
+
+template<typename _Tp> static inline
+std::ostream& operator << (std::ostream& out, const Rect_<_Tp>& rect)
+{
+    return out << "[" << rect.width << " x " << rect.height << " from (" << rect.x << ", " << rect.y << ")]";
+}
+
+
+#endif // OPENCV_NOSTL
+} // cv
+
+//! @endcond
+
+#endif // OPENCV_CORE_CVSTDINL_HPP

include/opencv2/core/directx.hpp

@@ -0,0 +1,184 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                           License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2010-2013, Advanced Micro Devices, Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors as is and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the copyright holders or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef OPENCV_CORE_DIRECTX_HPP
+#define OPENCV_CORE_DIRECTX_HPP
+
+#include "mat.hpp"
+#include "ocl.hpp"
+
+#if !defined(__d3d11_h__)
+struct ID3D11Device;
+struct ID3D11Texture2D;
+#endif
+
+#if !defined(__d3d10_h__)
+struct ID3D10Device;
+struct ID3D10Texture2D;
+#endif
+
+#if !defined(_D3D9_H_)
+struct IDirect3DDevice9;
+struct IDirect3DDevice9Ex;
+struct IDirect3DSurface9;
+#endif
+
+
+namespace cv { namespace directx {
+
+namespace ocl {
+using namespace cv::ocl;
+
+//! @addtogroup core_directx
+// This section describes OpenCL and DirectX interoperability.
+//
+// To enable DirectX support, configure OpenCV using CMake with WITH_DIRECTX=ON . Note, DirectX is
+// supported only on Windows.
+//
+// To use OpenCL functionality you should first initialize OpenCL context from DirectX resource.
+//
+//! @{
+
+// TODO static functions in the Context class
+//! @brief Creates OpenCL context from D3D11 device
+//
+//! @param pD3D11Device - pointer to D3D11 device
+//! @return Returns reference to OpenCL Context
+CV_EXPORTS Context& initializeContextFromD3D11Device(ID3D11Device* pD3D11Device);
+
+//! @brief Creates OpenCL context from D3D10 device
+//
+//! @param pD3D10Device - pointer to D3D10 device
+//! @return Returns reference to OpenCL Context
+CV_EXPORTS Context& initializeContextFromD3D10Device(ID3D10Device* pD3D10Device);
+
+//! @brief Creates OpenCL context from Direct3DDevice9Ex device
+//
+//! @param pDirect3DDevice9Ex - pointer to Direct3DDevice9Ex device
+//! @return Returns reference to OpenCL Context
+CV_EXPORTS Context& initializeContextFromDirect3DDevice9Ex(IDirect3DDevice9Ex* pDirect3DDevice9Ex);
+
+//! @brief Creates OpenCL context from Direct3DDevice9 device
+//
+//! @param pDirect3DDevice9 - pointer to Direct3Device9 device
+//! @return Returns reference to OpenCL Context
+CV_EXPORTS Context& initializeContextFromDirect3DDevice9(IDirect3DDevice9* pDirect3DDevice9);
+
+//! @}
+
+} // namespace cv::directx::ocl
+
+//! @addtogroup core_directx
+//! @{
+
+//! @brief Converts InputArray to ID3D11Texture2D. If destination texture format is DXGI_FORMAT_NV12 then
+//!        input UMat expected to be in BGR format and data will be downsampled and color-converted to NV12.
+//
+//! @note Note: Destination texture must be allocated by application. Function does memory copy from src to
+//!             pD3D11Texture2D
+//
+//! @param src - source InputArray
+//! @param pD3D11Texture2D - destination D3D11 texture
+CV_EXPORTS void convertToD3D11Texture2D(InputArray src, ID3D11Texture2D* pD3D11Texture2D);
+
+//! @brief Converts ID3D11Texture2D to OutputArray. If input texture format is DXGI_FORMAT_NV12 then
+//!        data will be upsampled and color-converted to BGR format.
+//
+//! @note Note: Destination matrix will be re-allocated if it has not enough memory to match texture size.
+//!             function does memory copy from pD3D11Texture2D to dst
+//
+//! @param pD3D11Texture2D - source D3D11 texture
+//! @param dst             - destination OutputArray
+CV_EXPORTS void convertFromD3D11Texture2D(ID3D11Texture2D* pD3D11Texture2D, OutputArray dst);
+
+//! @brief Converts InputArray to ID3D10Texture2D
+//
+//! @note Note: function does memory copy from src to
+//!             pD3D10Texture2D
+//
+//! @param src             - source InputArray
+//! @param pD3D10Texture2D - destination D3D10 texture
+CV_EXPORTS void convertToD3D10Texture2D(InputArray src, ID3D10Texture2D* pD3D10Texture2D);
+
+//! @brief Converts ID3D10Texture2D to OutputArray
+//
+//! @note Note: function does memory copy from pD3D10Texture2D
+//!             to dst
+//
+//! @param pD3D10Texture2D - source D3D10 texture
+//! @param dst             - destination OutputArray
+CV_EXPORTS void convertFromD3D10Texture2D(ID3D10Texture2D* pD3D10Texture2D, OutputArray dst);
+
+//! @brief Converts InputArray to IDirect3DSurface9
+//
+//! @note Note: function does memory copy from src to
+//!             pDirect3DSurface9
+//
+//! @param src                 - source InputArray
+//! @param pDirect3DSurface9   - destination D3D10 texture
+//! @param surfaceSharedHandle - shared handle
+CV_EXPORTS void convertToDirect3DSurface9(InputArray src, IDirect3DSurface9* pDirect3DSurface9, void* surfaceSharedHandle = NULL);
+
+//! @brief Converts IDirect3DSurface9 to OutputArray
+//
+//! @note Note: function does memory copy from pDirect3DSurface9
+//!             to dst
+//
+//! @param pDirect3DSurface9   - source D3D10 texture
+//! @param dst                 - destination OutputArray
+//! @param surfaceSharedHandle - shared handle
+CV_EXPORTS void convertFromDirect3DSurface9(IDirect3DSurface9* pDirect3DSurface9, OutputArray dst, void* surfaceSharedHandle = NULL);
+
+//! @brief Get OpenCV type from DirectX type
+//! @param iDXGI_FORMAT - enum DXGI_FORMAT for D3D10/D3D11
+//! @return OpenCV type or -1 if there is no equivalent
+CV_EXPORTS int getTypeFromDXGI_FORMAT(const int iDXGI_FORMAT); // enum DXGI_FORMAT for D3D10/D3D11
+
+//! @brief Get OpenCV type from DirectX type
+//! @param iD3DFORMAT - enum D3DTYPE for D3D9
+//! @return OpenCV type or -1 if there is no equivalent
+CV_EXPORTS int getTypeFromD3DFORMAT(const int iD3DFORMAT); // enum D3DTYPE for D3D9
+
+//! @}
+
+} } // namespace cv::directx
+
+#endif // OPENCV_CORE_DIRECTX_HPP

include/opencv2/core/eigen.hpp

@@ -12,6 +12,7 @@
 //
 // Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
 // Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
 // Third party copyrights are property of their respective owners.
 //
 // Redistribution and use in source and binary forms, with or without modification,
@@ -40,13 +41,11 @@
 //
 //M*/
 
-#ifndef __OPENCV_CORE_EIGEN_HPP__
-#define __OPENCV_CORE_EIGEN_HPP__
 
-#ifdef __cplusplus
+#ifndef OPENCV_CORE_EIGEN_HPP
+#define OPENCV_CORE_EIGEN_HPP
 
-#include "opencv2/core/core_c.h"
-#include "opencv2/core/core.hpp"
+#include "opencv2/core.hpp"
 
 #if defined _MSC_VER && _MSC_VER >= 1200
 #pragma warning( disable: 4714 ) //__forceinline is not inlined
@@ -57,7 +56,10 @@
 namespace cv
 {
 
-template<typename _Tp, int _rows, int _cols, int _options, int _maxRows, int _maxCols>
+//! @addtogroup core_eigen
+//! @{
+
+template<typename _Tp, int _rows, int _cols, int _options, int _maxRows, int _maxCols> static inline
 void eigen2cv( const Eigen::Matrix<_Tp, _rows, _cols, _options, _maxRows, _maxCols>& src, Mat& dst )
 {
     if( !(src.Flags & Eigen::RowMajorBit) )
@@ -74,14 +76,29 @@ void eigen2cv( const Eigen::Matrix<_Tp, _rows, _cols, _options, _maxRows, _maxCo
     }
 }
 
-template<typename _Tp, int _rows, int _cols, int _options, int _maxRows, int _maxCols>
+// Matx case
+template<typename _Tp, int _rows, int _cols, int _options, int _maxRows, int _maxCols> static inline
+void eigen2cv( const Eigen::Matrix<_Tp, _rows, _cols, _options, _maxRows, _maxCols>& src,
+               Matx<_Tp, _rows, _cols>& dst )
+{
+    if( !(src.Flags & Eigen::RowMajorBit) )
+    {
+        dst = Matx<_Tp, _cols, _rows>(static_cast<const _Tp*>(src.data())).t();
+    }
+    else
+    {
+        dst = Matx<_Tp, _rows, _cols>(static_cast<const _Tp*>(src.data()));
+    }
+}
+
+template<typename _Tp, int _rows, int _cols, int _options, int _maxRows, int _maxCols> static inline
 void cv2eigen( const Mat& src,
                Eigen::Matrix<_Tp, _rows, _cols, _options, _maxRows, _maxCols>& dst )
 {
     CV_DbgAssert(src.rows == _rows && src.cols == _cols);
     if( !(dst.Flags & Eigen::RowMajorBit) )
     {
-        Mat _dst(src.cols, src.rows, DataType<_Tp>::type,
+        const Mat _dst(src.cols, src.rows, DataType<_Tp>::type,
                  dst.data(), (size_t)(dst.stride()*sizeof(_Tp)));
         if( src.type() == _dst.type() )
             transpose(src, _dst);
@@ -92,46 +109,42 @@ void cv2eigen( const Mat& src,
         }
         else
             Mat(src.t()).convertTo(_dst, _dst.type());
-        CV_DbgAssert(_dst.data == (uchar*)dst.data());
     }
     else
     {
-        Mat _dst(src.rows, src.cols, DataType<_Tp>::type,
+        const Mat _dst(src.rows, src.cols, DataType<_Tp>::type,
                  dst.data(), (size_t)(dst.stride()*sizeof(_Tp)));
         src.convertTo(_dst, _dst.type());
-        CV_DbgAssert(_dst.data == (uchar*)dst.data());
     }
 }
 
 // Matx case
-template<typename _Tp, int _rows, int _cols, int _options, int _maxRows, int _maxCols>
+template<typename _Tp, int _rows, int _cols, int _options, int _maxRows, int _maxCols> static inline
 void cv2eigen( const Matx<_Tp, _rows, _cols>& src,
                Eigen::Matrix<_Tp, _rows, _cols, _options, _maxRows, _maxCols>& dst )
 {
     if( !(dst.Flags & Eigen::RowMajorBit) )
     {
-        Mat _dst(_cols, _rows, DataType<_Tp>::type,
+        const Mat _dst(_cols, _rows, DataType<_Tp>::type,
                  dst.data(), (size_t)(dst.stride()*sizeof(_Tp)));
         transpose(src, _dst);
-        CV_DbgAssert(_dst.data == (uchar*)dst.data());
     }
     else
     {
-        Mat _dst(_rows, _cols, DataType<_Tp>::type,
+        const Mat _dst(_rows, _cols, DataType<_Tp>::type,
                  dst.data(), (size_t)(dst.stride()*sizeof(_Tp)));
         Mat(src).copyTo(_dst);
-        CV_DbgAssert(_dst.data == (uchar*)dst.data());
     }
 }
 
-template<typename _Tp>
+template<typename _Tp>  static inline
 void cv2eigen( const Mat& src,
                Eigen::Matrix<_Tp, Eigen::Dynamic, Eigen::Dynamic>& dst )
 {
     dst.resize(src.rows, src.cols);
     if( !(dst.Flags & Eigen::RowMajorBit) )
     {
-        Mat _dst(src.cols, src.rows, DataType<_Tp>::type,
+        const Mat _dst(src.cols, src.rows, DataType<_Tp>::type,
              dst.data(), (size_t)(dst.stride()*sizeof(_Tp)));
         if( src.type() == _dst.type() )
             transpose(src, _dst);
@@ -142,40 +155,36 @@ void cv2eigen( const Mat& src,
         }
         else
             Mat(src.t()).convertTo(_dst, _dst.type());
-        CV_DbgAssert(_dst.data == (uchar*)dst.data());
     }
     else
     {
-        Mat _dst(src.rows, src.cols, DataType<_Tp>::type,
+        const Mat _dst(src.rows, src.cols, DataType<_Tp>::type,
                  dst.data(), (size_t)(dst.stride()*sizeof(_Tp)));
         src.convertTo(_dst, _dst.type());
-        CV_DbgAssert(_dst.data == (uchar*)dst.data());
     }
 }
 
 // Matx case
-template<typename _Tp, int _rows, int _cols>
+template<typename _Tp, int _rows, int _cols> static inline
 void cv2eigen( const Matx<_Tp, _rows, _cols>& src,
                Eigen::Matrix<_Tp, Eigen::Dynamic, Eigen::Dynamic>& dst )
 {
     dst.resize(_rows, _cols);
     if( !(dst.Flags & Eigen::RowMajorBit) )
     {
-        Mat _dst(_cols, _rows, DataType<_Tp>::type,
+        const Mat _dst(_cols, _rows, DataType<_Tp>::type,
              dst.data(), (size_t)(dst.stride()*sizeof(_Tp)));
         transpose(src, _dst);
-        CV_DbgAssert(_dst.data == (uchar*)dst.data());
     }
     else
     {
-        Mat _dst(_rows, _cols, DataType<_Tp>::type,
+        const Mat _dst(_rows, _cols, DataType<_Tp>::type,
                  dst.data(), (size_t)(dst.stride()*sizeof(_Tp)));
         Mat(src).copyTo(_dst);
-        CV_DbgAssert(_dst.data == (uchar*)dst.data());
     }
 }
 
-template<typename _Tp>
+template<typename _Tp> static inline
 void cv2eigen( const Mat& src,
                Eigen::Matrix<_Tp, Eigen::Dynamic, 1>& dst )
 {
@@ -184,25 +193,23 @@ void cv2eigen( const Mat& src,
 
     if( !(dst.Flags & Eigen::RowMajorBit) )
     {
-        Mat _dst(src.cols, src.rows, DataType<_Tp>::type,
+        const Mat _dst(src.cols, src.rows, DataType<_Tp>::type,
                  dst.data(), (size_t)(dst.stride()*sizeof(_Tp)));
         if( src.type() == _dst.type() )
             transpose(src, _dst);
         else
             Mat(src.t()).convertTo(_dst, _dst.type());
-        CV_DbgAssert(_dst.data == (uchar*)dst.data());
     }
     else
     {
-        Mat _dst(src.rows, src.cols, DataType<_Tp>::type,
+        const Mat _dst(src.rows, src.cols, DataType<_Tp>::type,
                  dst.data(), (size_t)(dst.stride()*sizeof(_Tp)));
         src.convertTo(_dst, _dst.type());
-        CV_DbgAssert(_dst.data == (uchar*)dst.data());
     }
 }
 
 // Matx case
-template<typename _Tp, int _rows>
+template<typename _Tp, int _rows> static inline
 void cv2eigen( const Matx<_Tp, _rows, 1>& src,
                Eigen::Matrix<_Tp, Eigen::Dynamic, 1>& dst )
 {
@@ -210,22 +217,20 @@ void cv2eigen( const Matx<_Tp, _rows, 1>& src,
 
     if( !(dst.Flags & Eigen::RowMajorBit) )
     {
-        Mat _dst(1, _rows, DataType<_Tp>::type,
+        const Mat _dst(1, _rows, DataType<_Tp>::type,
                  dst.data(), (size_t)(dst.stride()*sizeof(_Tp)));
         transpose(src, _dst);
-        CV_DbgAssert(_dst.data == (uchar*)dst.data());
     }
     else
     {
-        Mat _dst(_rows, 1, DataType<_Tp>::type,
+        const Mat _dst(_rows, 1, DataType<_Tp>::type,
                  dst.data(), (size_t)(dst.stride()*sizeof(_Tp)));
         src.copyTo(_dst);
-        CV_DbgAssert(_dst.data == (uchar*)dst.data());
     }
 }
 
 
-template<typename _Tp>
+template<typename _Tp> static inline
 void cv2eigen( const Mat& src,
                Eigen::Matrix<_Tp, 1, Eigen::Dynamic>& dst )
 {
@@ -233,48 +238,43 @@ void cv2eigen( const Mat& src,
     dst.resize(src.cols);
     if( !(dst.Flags & Eigen::RowMajorBit) )
     {
-        Mat _dst(src.cols, src.rows, DataType<_Tp>::type,
+        const Mat _dst(src.cols, src.rows, DataType<_Tp>::type,
                  dst.data(), (size_t)(dst.stride()*sizeof(_Tp)));
         if( src.type() == _dst.type() )
             transpose(src, _dst);
         else
             Mat(src.t()).convertTo(_dst, _dst.type());
-        CV_DbgAssert(_dst.data == (uchar*)dst.data());
     }
     else
     {
-        Mat _dst(src.rows, src.cols, DataType<_Tp>::type,
+        const Mat _dst(src.rows, src.cols, DataType<_Tp>::type,
                  dst.data(), (size_t)(dst.stride()*sizeof(_Tp)));
         src.convertTo(_dst, _dst.type());
-        CV_DbgAssert(_dst.data == (uchar*)dst.data());
     }
 }
 
 //Matx
-template<typename _Tp, int _cols>
+template<typename _Tp, int _cols> static inline
 void cv2eigen( const Matx<_Tp, 1, _cols>& src,
                Eigen::Matrix<_Tp, 1, Eigen::Dynamic>& dst )
 {
     dst.resize(_cols);
     if( !(dst.Flags & Eigen::RowMajorBit) )
     {
-        Mat _dst(_cols, 1, DataType<_Tp>::type,
+        const Mat _dst(_cols, 1, DataType<_Tp>::type,
                  dst.data(), (size_t)(dst.stride()*sizeof(_Tp)));
         transpose(src, _dst);
-        CV_DbgAssert(_dst.data == (uchar*)dst.data());
     }
     else
     {
-        Mat _dst(1, _cols, DataType<_Tp>::type,
+        const Mat _dst(1, _cols, DataType<_Tp>::type,
                  dst.data(), (size_t)(dst.stride()*sizeof(_Tp)));
         Mat(src).copyTo(_dst);
-        CV_DbgAssert(_dst.data == (uchar*)dst.data());
     }
 }
 
+//! @}
 
-}
-
-#endif
+} // cv
 
 #endif

include/opencv2/core/fast_math.hpp

@@ -0,0 +1,303 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                          License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
+// Copyright (C) 2015, Itseez Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef OPENCV_CORE_FAST_MATH_HPP
+#define OPENCV_CORE_FAST_MATH_HPP
+
+#include "opencv2/core/cvdef.h"
+
+//! @addtogroup core_utils
+//! @{
+
+/****************************************************************************************\
+*                                      fast math                                         *
+\****************************************************************************************/
+
+#if defined __BORLANDC__
+#  include <fastmath.h>
+#elif defined __cplusplus
+#  include <cmath>
+#else
+#  include <math.h>
+#endif
+
+#ifdef HAVE_TEGRA_OPTIMIZATION
+#  include "tegra_round.hpp"
+#endif
+
+#if CV_VFP
+    // 1. general scheme
+    #define ARM_ROUND(_value, _asm_string) \
+        int res; \
+        float temp; \
+        (void)temp; \
+        asm(_asm_string : [res] "=r" (res), [temp] "=w" (temp) : [value] "w" (_value)); \
+        return res
+    // 2. version for double
+    #ifdef __clang__
+        #define ARM_ROUND_DBL(value) ARM_ROUND(value, "vcvtr.s32.f64 %[temp], %[value] \n vmov %[res], %[temp]")
+    #else
+        #define ARM_ROUND_DBL(value) ARM_ROUND(value, "vcvtr.s32.f64 %[temp], %P[value] \n vmov %[res], %[temp]")
+    #endif
+    // 3. version for float
+    #define ARM_ROUND_FLT(value) ARM_ROUND(value, "vcvtr.s32.f32 %[temp], %[value]\n vmov %[res], %[temp]")
+#endif // CV_VFP
+
+/** @brief Rounds floating-point number to the nearest integer
+
+ @param value floating-point number. If the value is outside of INT_MIN ... INT_MAX range, the
+ result is not defined.
+ */
+CV_INLINE int
+cvRound( double value )
+{
+#if ((defined _MSC_VER && defined _M_X64) || (defined __GNUC__ && defined __x86_64__ \
+    && defined __SSE2__ && !defined __APPLE__)) && !defined(__CUDACC__)
+    __m128d t = _mm_set_sd( value );
+    return _mm_cvtsd_si32(t);
+#elif defined _MSC_VER && defined _M_IX86
+    int t;
+    __asm
+    {
+        fld value;
+        fistp t;
+    }
+    return t;
+#elif ((defined _MSC_VER && defined _M_ARM) || defined CV_ICC || \
+        defined __GNUC__) && defined HAVE_TEGRA_OPTIMIZATION
+    TEGRA_ROUND_DBL(value);
+#elif defined CV_ICC || defined __GNUC__
+# if CV_VFP
+    ARM_ROUND_DBL(value);
+# else
+    return (int)lrint(value);
+# endif
+#else
+    /* it's ok if round does not comply with IEEE754 standard;
+       the tests should allow +/-1 difference when the tested functions use round */
+    return (int)(value + (value >= 0 ? 0.5 : -0.5));
+#endif
+}
+
+
+/** @brief Rounds floating-point number to the nearest integer not larger than the original.
+
+ The function computes an integer i such that:
+ \f[i \le \texttt{value} < i+1\f]
+ @param value floating-point number. If the value is outside of INT_MIN ... INT_MAX range, the
+ result is not defined.
+ */
+CV_INLINE int cvFloor( double value )
+{
+#if (defined _MSC_VER && defined _M_X64 || (defined __GNUC__ && defined __SSE2__ && !defined __APPLE__)) && !defined(__CUDACC__)
+    __m128d t = _mm_set_sd( value );
+    int i = _mm_cvtsd_si32(t);
+    return i - _mm_movemask_pd(_mm_cmplt_sd(t, _mm_cvtsi32_sd(t,i)));
+#elif defined __GNUC__
+    int i = (int)value;
+    return i - (i > value);
+#else
+    int i = cvRound(value);
+    float diff = (float)(value - i);
+    return i - (diff < 0);
+#endif
+}
+
+/** @brief Rounds floating-point number to the nearest integer not smaller than the original.
+
+ The function computes an integer i such that:
+ \f[i \le \texttt{value} < i+1\f]
+ @param value floating-point number. If the value is outside of INT_MIN ... INT_MAX range, the
+ result is not defined.
+ */
+CV_INLINE int cvCeil( double value )
+{
+#if (defined _MSC_VER && defined _M_X64 || (defined __GNUC__ && defined __SSE2__&& !defined __APPLE__)) && !defined(__CUDACC__)
+    __m128d t = _mm_set_sd( value );
+    int i = _mm_cvtsd_si32(t);
+    return i + _mm_movemask_pd(_mm_cmplt_sd(_mm_cvtsi32_sd(t,i), t));
+#elif defined __GNUC__
+    int i = (int)value;
+    return i + (i < value);
+#else
+    int i = cvRound(value);
+    float diff = (float)(i - value);
+    return i + (diff < 0);
+#endif
+}
+
+/** @brief Determines if the argument is Not A Number.
+
+ @param value The input floating-point value
+
+ The function returns 1 if the argument is Not A Number (as defined by IEEE754 standard), 0
+ otherwise. */
+CV_INLINE int cvIsNaN( double value )
+{
+    Cv64suf ieee754;
+    ieee754.f = value;
+    return ((unsigned)(ieee754.u >> 32) & 0x7fffffff) +
+           ((unsigned)ieee754.u != 0) > 0x7ff00000;
+}
+
+/** @brief Determines if the argument is Infinity.
+
+ @param value The input floating-point value
+
+ The function returns 1 if the argument is a plus or minus infinity (as defined by IEEE754 standard)
+ and 0 otherwise. */
+CV_INLINE int cvIsInf( double value )
+{
+    Cv64suf ieee754;
+    ieee754.f = value;
+    return ((unsigned)(ieee754.u >> 32) & 0x7fffffff) == 0x7ff00000 &&
+            (unsigned)ieee754.u == 0;
+}
+
+#ifdef __cplusplus
+
+/** @overload */
+CV_INLINE int cvRound(float value)
+{
+#if ((defined _MSC_VER && defined _M_X64) || (defined __GNUC__ && defined __x86_64__ && \
+      defined __SSE2__ && !defined __APPLE__)) && !defined(__CUDACC__)
+    __m128 t = _mm_set_ss( value );
+    return _mm_cvtss_si32(t);
+#elif defined _MSC_VER && defined _M_IX86
+    int t;
+    __asm
+    {
+        fld value;
+        fistp t;
+    }
+    return t;
+#elif ((defined _MSC_VER && defined _M_ARM) || defined CV_ICC || \
+        defined __GNUC__) && defined HAVE_TEGRA_OPTIMIZATION
+    TEGRA_ROUND_FLT(value);
+#elif defined CV_ICC || defined __GNUC__
+# if CV_VFP
+    ARM_ROUND_FLT(value);
+# else
+    return (int)lrintf(value);
+# endif
+#else
+    /* it's ok if round does not comply with IEEE754 standard;
+     the tests should allow +/-1 difference when the tested functions use round */
+    return (int)(value + (value >= 0 ? 0.5f : -0.5f));
+#endif
+}
+
+/** @overload */
+CV_INLINE int cvRound( int value )
+{
+    return value;
+}
+
+/** @overload */
+CV_INLINE int cvFloor( float value )
+{
+#if (defined _MSC_VER && defined _M_X64 || (defined __GNUC__ && defined __SSE2__ && !defined __APPLE__)) && !defined(__CUDACC__)
+    __m128 t = _mm_set_ss( value );
+    int i = _mm_cvtss_si32(t);
+    return i - _mm_movemask_ps(_mm_cmplt_ss(t, _mm_cvtsi32_ss(t,i)));
+#elif defined __GNUC__
+    int i = (int)value;
+    return i - (i > value);
+#else
+    int i = cvRound(value);
+    float diff = (float)(value - i);
+    return i - (diff < 0);
+#endif
+}
+
+/** @overload */
+CV_INLINE int cvFloor( int value )
+{
+    return value;
+}
+
+/** @overload */
+CV_INLINE int cvCeil( float value )
+{
+#if (defined _MSC_VER && defined _M_X64 || (defined __GNUC__ && defined __SSE2__&& !defined __APPLE__)) && !defined(__CUDACC__)
+    __m128 t = _mm_set_ss( value );
+    int i = _mm_cvtss_si32(t);
+    return i + _mm_movemask_ps(_mm_cmplt_ss(_mm_cvtsi32_ss(t,i), t));
+#elif defined __GNUC__
+    int i = (int)value;
+    return i + (i < value);
+#else
+    int i = cvRound(value);
+    float diff = (float)(i - value);
+    return i + (diff < 0);
+#endif
+}
+
+/** @overload */
+CV_INLINE int cvCeil( int value )
+{
+    return value;
+}
+
+/** @overload */
+CV_INLINE int cvIsNaN( float value )
+{
+    Cv32suf ieee754;
+    ieee754.f = value;
+    return (ieee754.u & 0x7fffffff) > 0x7f800000;
+}
+
+/** @overload */
+CV_INLINE int cvIsInf( float value )
+{
+    Cv32suf ieee754;
+    ieee754.f = value;
+    return (ieee754.u & 0x7fffffff) == 0x7f800000;
+}
+
+#endif // __cplusplus
+
+//! @} core_utils
+
+#endif

include/opencv2/core/gpumat.hpp

@@ -1,562 +0,0 @@
-/*M///////////////////////////////////////////////////////////////////////////////////////
-//
-//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
-//
-//  By downloading, copying, installing or using the software you agree to this license.
-//  If you do not agree to this license, do not download, install,
-//  copy or use the software.
-//
-//
-//                           License Agreement
-//                For Open Source Computer Vision Library
-//
-// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
-// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
-// Third party copyrights are property of their respective owners.
-//
-// Redistribution and use in source and binary forms, with or without modification,
-// are permitted provided that the following conditions are met:
-//
-//   * Redistribution's of source code must retain the above copyright notice,
-//     this list of conditions and the following disclaimer.
-//
-//   * Redistribution's in binary form must reproduce the above copyright notice,
-//     this list of conditions and the following disclaimer in the documentation
-//     and/or other materials provided with the distribution.
-//
-//   * The name of the copyright holders may not be used to endorse or promote products
-//     derived from this software without specific prior written permission.
-//
-// This software is provided by the copyright holders and contributors "as is" and
-// any express or implied warranties, including, but not limited to, the implied
-// warranties of merchantability and fitness for a particular purpose are disclaimed.
-// In no event shall the Intel Corporation or contributors be liable for any direct,
-// indirect, incidental, special, exemplary, or consequential damages
-// (including, but not limited to, procurement of substitute goods or services;
-// loss of use, data, or profits; or business interruption) however caused
-// and on any theory of liability, whether in contract, strict liability,
-// or tort (including negligence or otherwise) arising in any way out of
-// the use of this software, even if advised of the possibility of such damage.
-//
-//M*/
-
-#ifndef __OPENCV_GPUMAT_HPP__
-#define __OPENCV_GPUMAT_HPP__
-
-#ifdef __cplusplus
-
-#include "opencv2/core/core.hpp"
-#include "opencv2/core/cuda_devptrs.hpp"
-
-namespace cv { namespace gpu
-{
-    //////////////////////////////// Initialization & Info ////////////////////////
-
-    //! This is the only function that do not throw exceptions if the library is compiled without Cuda.
-    CV_EXPORTS int getCudaEnabledDeviceCount();
-
-    //! Functions below throw cv::Expception if the library is compiled without Cuda.
-
-    CV_EXPORTS void setDevice(int device);
-    CV_EXPORTS int getDevice();
-
-    //! Explicitly destroys and cleans up all resources associated with the current device in the current process.
-    //! Any subsequent API call to this device will reinitialize the device.
-    CV_EXPORTS void resetDevice();
-
-    enum FeatureSet
-    {
-        FEATURE_SET_COMPUTE_10 = 10,
-        FEATURE_SET_COMPUTE_11 = 11,
-        FEATURE_SET_COMPUTE_12 = 12,
-        FEATURE_SET_COMPUTE_13 = 13,
-        FEATURE_SET_COMPUTE_20 = 20,
-        FEATURE_SET_COMPUTE_21 = 21,
-        FEATURE_SET_COMPUTE_30 = 30,
-        FEATURE_SET_COMPUTE_35 = 35,
-
-        GLOBAL_ATOMICS = FEATURE_SET_COMPUTE_11,
-        SHARED_ATOMICS = FEATURE_SET_COMPUTE_12,
-        NATIVE_DOUBLE = FEATURE_SET_COMPUTE_13,
-        WARP_SHUFFLE_FUNCTIONS = FEATURE_SET_COMPUTE_30,
-        DYNAMIC_PARALLELISM = FEATURE_SET_COMPUTE_35
-    };
-
-    // Checks whether current device supports the given feature
-    CV_EXPORTS bool deviceSupports(FeatureSet feature_set);
-
-    // Gives information about what GPU archs this OpenCV GPU module was
-    // compiled for
-    class CV_EXPORTS TargetArchs
-    {
-    public:
-        static bool builtWith(FeatureSet feature_set);
-        static bool has(int major, int minor);
-        static bool hasPtx(int major, int minor);
-        static bool hasBin(int major, int minor);
-        static bool hasEqualOrLessPtx(int major, int minor);
-        static bool hasEqualOrGreater(int major, int minor);
-        static bool hasEqualOrGreaterPtx(int major, int minor);
-        static bool hasEqualOrGreaterBin(int major, int minor);
-    private:
-        TargetArchs();
-    };
-
-    // Gives information about the given GPU
-    class CV_EXPORTS DeviceInfo
-    {
-    public:
-        // Creates DeviceInfo object for the current GPU
-        DeviceInfo() : device_id_(getDevice()) { query(); }
-
-        // Creates DeviceInfo object for the given GPU
-        DeviceInfo(int device_id) : device_id_(device_id) { query(); }
-
-        std::string name() const { return name_; }
-
-        // Return compute capability versions
-        int majorVersion() const { return majorVersion_; }
-        int minorVersion() const { return minorVersion_; }
-
-        int multiProcessorCount() const { return multi_processor_count_; }
-
-        size_t sharedMemPerBlock() const;
-
-        void queryMemory(size_t& totalMemory, size_t& freeMemory) const;
-        size_t freeMemory() const;
-        size_t totalMemory() const;
-
-        // Checks whether device supports the given feature
-        bool supports(FeatureSet feature_set) const;
-
-        // Checks whether the GPU module can be run on the given device
-        bool isCompatible() const;
-
-        int deviceID() const { return device_id_; }
-
-    private:
-        void query();
-
-        int device_id_;
-
-        std::string name_;
-        int multi_processor_count_;
-        int majorVersion_;
-        int minorVersion_;
-    };
-
-    CV_EXPORTS void printCudaDeviceInfo(int device);
-    CV_EXPORTS void printShortCudaDeviceInfo(int device);
-
-    //////////////////////////////// GpuMat ///////////////////////////////
-
-    //! Smart pointer for GPU memory with reference counting. Its interface is mostly similar with cv::Mat.
-    class CV_EXPORTS GpuMat
-    {
-    public:
-        //! default constructor
-        GpuMat();
-
-        //! constructs GpuMatrix of the specified size and type (_type is CV_8UC1, CV_64FC3, CV_32SC(12) etc.)
-        GpuMat(int rows, int cols, int type);
-        GpuMat(Size size, int type);
-
-        //! constucts GpuMatrix and fills it with the specified value _s.
-        GpuMat(int rows, int cols, int type, Scalar s);
-        GpuMat(Size size, int type, Scalar s);
-
-        //! copy constructor
-        GpuMat(const GpuMat& m);
-
-        //! constructor for GpuMatrix headers pointing to user-allocated data
-        GpuMat(int rows, int cols, int type, void* data, size_t step = Mat::AUTO_STEP);
-        GpuMat(Size size, int type, void* data, size_t step = Mat::AUTO_STEP);
-
-        //! creates a matrix header for a part of the bigger matrix
-        GpuMat(const GpuMat& m, Range rowRange, Range colRange);
-        GpuMat(const GpuMat& m, Rect roi);
-
-        //! builds GpuMat from Mat. Perfom blocking upload to device.
-        explicit GpuMat(const Mat& m);
-
-        //! destructor - calls release()
-        ~GpuMat();
-
-        //! assignment operators
-        GpuMat& operator = (const GpuMat& m);
-
-        //! pefroms blocking upload data to GpuMat.
-        void upload(const Mat& m);
-
-        //! downloads data from device to host memory. Blocking calls.
-        void download(Mat& m) const;
-
-        //! returns a new GpuMatrix header for the specified row
-        GpuMat row(int y) const;
-        //! returns a new GpuMatrix header for the specified column
-        GpuMat col(int x) const;
-        //! ... for the specified row span
-        GpuMat rowRange(int startrow, int endrow) const;
-        GpuMat rowRange(Range r) const;
-        //! ... for the specified column span
-        GpuMat colRange(int startcol, int endcol) const;
-        GpuMat colRange(Range r) const;
-
-        //! returns deep copy of the GpuMatrix, i.e. the data is copied
-        GpuMat clone() const;
-        //! copies the GpuMatrix content to "m".
-        // It calls m.create(this->size(), this->type()).
-        void copyTo(GpuMat& m) const;
-        //! copies those GpuMatrix elements to "m" that are marked with non-zero mask elements.
-        void copyTo(GpuMat& m, const GpuMat& mask) const;
-        //! converts GpuMatrix to another datatype with optional scalng. See cvConvertScale.
-        void convertTo(GpuMat& m, int rtype, double alpha = 1, double beta = 0) const;
-
-        void assignTo(GpuMat& m, int type=-1) const;
-
-        //! sets every GpuMatrix element to s
-        GpuMat& operator = (Scalar s);
-        //! sets some of the GpuMatrix elements to s, according to the mask
-        GpuMat& setTo(Scalar s, const GpuMat& mask = GpuMat());
-        //! creates alternative GpuMatrix header for the same data, with different
-        // number of channels and/or different number of rows. see cvReshape.
-        GpuMat reshape(int cn, int rows = 0) const;
-
-        //! allocates new GpuMatrix data unless the GpuMatrix already has specified size and type.
-        // previous data is unreferenced if needed.
-        void create(int rows, int cols, int type);
-        void create(Size size, int type);
-        //! decreases reference counter;
-        // deallocate the data when reference counter reaches 0.
-        void release();
-
-        //! swaps with other smart pointer
-        void swap(GpuMat& mat);
-
-        //! locates GpuMatrix header within a parent GpuMatrix. See below
-        void locateROI(Size& wholeSize, Point& ofs) const;
-        //! moves/resizes the current GpuMatrix ROI inside the parent GpuMatrix.
-        GpuMat& adjustROI(int dtop, int dbottom, int dleft, int dright);
-        //! extracts a rectangular sub-GpuMatrix
-        // (this is a generalized form of row, rowRange etc.)
-        GpuMat operator()(Range rowRange, Range colRange) const;
-        GpuMat operator()(Rect roi) const;
-
-        //! returns true iff the GpuMatrix data is continuous
-        // (i.e. when there are no gaps between successive rows).
-        // similar to CV_IS_GpuMat_CONT(cvGpuMat->type)
-        bool isContinuous() const;
-        //! returns element size in bytes,
-        // similar to CV_ELEM_SIZE(cvMat->type)
-        size_t elemSize() const;
-        //! returns the size of element channel in bytes.
-        size_t elemSize1() const;
-        //! returns element type, similar to CV_MAT_TYPE(cvMat->type)
-        int type() const;
-        //! returns element type, similar to CV_MAT_DEPTH(cvMat->type)
-        int depth() const;
-        //! returns element type, similar to CV_MAT_CN(cvMat->type)
-        int channels() const;
-        //! returns step/elemSize1()
-        size_t step1() const;
-        //! returns GpuMatrix size:
-        // width == number of columns, height == number of rows
-        Size size() const;
-        //! returns true if GpuMatrix data is NULL
-        bool empty() const;
-
-        //! returns pointer to y-th row
-        uchar* ptr(int y = 0);
-        const uchar* ptr(int y = 0) const;
-
-        //! template version of the above method
-        template<typename _Tp> _Tp* ptr(int y = 0);
-        template<typename _Tp> const _Tp* ptr(int y = 0) const;
-
-        template <typename _Tp> operator PtrStepSz<_Tp>() const;
-        template <typename _Tp> operator PtrStep<_Tp>() const;
-
-        // Deprecated function
-        __CV_GPU_DEPR_BEFORE__ template <typename _Tp> operator DevMem2D_<_Tp>() const __CV_GPU_DEPR_AFTER__;
-        __CV_GPU_DEPR_BEFORE__ template <typename _Tp> operator PtrStep_<_Tp>() const __CV_GPU_DEPR_AFTER__;
-        #undef __CV_GPU_DEPR_BEFORE__
-        #undef __CV_GPU_DEPR_AFTER__
-
-        /*! includes several bit-fields:
-        - the magic signature
-        - continuity flag
-        - depth
-        - number of channels
-        */
-        int flags;
-
-        //! the number of rows and columns
-        int rows, cols;
-
-        //! a distance between successive rows in bytes; includes the gap if any
-        size_t step;
-
-        //! pointer to the data
-        uchar* data;
-
-        //! pointer to the reference counter;
-        // when GpuMatrix points to user-allocated data, the pointer is NULL
-        int* refcount;
-
-        //! helper fields used in locateROI and adjustROI
-        uchar* datastart;
-        uchar* dataend;
-    };
-
-    //! Creates continuous GPU matrix
-    CV_EXPORTS void createContinuous(int rows, int cols, int type, GpuMat& m);
-    CV_EXPORTS GpuMat createContinuous(int rows, int cols, int type);
-    CV_EXPORTS void createContinuous(Size size, int type, GpuMat& m);
-    CV_EXPORTS GpuMat createContinuous(Size size, int type);
-
-    //! Ensures that size of the given matrix is not less than (rows, cols) size
-    //! and matrix type is match specified one too
-    CV_EXPORTS void ensureSizeIsEnough(int rows, int cols, int type, GpuMat& m);
-    CV_EXPORTS void ensureSizeIsEnough(Size size, int type, GpuMat& m);
-
-    CV_EXPORTS GpuMat allocMatFromBuf(int rows, int cols, int type, GpuMat &mat);
-
-    ////////////////////////////////////////////////////////////////////////
-    // Error handling
-
-    CV_EXPORTS void error(const char* error_string, const char* file, const int line, const char* func = "");
-
-    ////////////////////////////////////////////////////////////////////////
-    ////////////////////////////////////////////////////////////////////////
-    ////////////////////////////////////////////////////////////////////////
-
-    inline GpuMat::GpuMat()
-        : flags(0), rows(0), cols(0), step(0), data(0), refcount(0), datastart(0), dataend(0)
-    {
-    }
-
-    inline GpuMat::GpuMat(int rows_, int cols_, int type_)
-        : flags(0), rows(0), cols(0), step(0), data(0), refcount(0), datastart(0), dataend(0)
-    {
-        if (rows_ > 0 && cols_ > 0)
-            create(rows_, cols_, type_);
-    }
-
-    inline GpuMat::GpuMat(Size size_, int type_)
-        : flags(0), rows(0), cols(0), step(0), data(0), refcount(0), datastart(0), dataend(0)
-    {
-        if (size_.height > 0 && size_.width > 0)
-            create(size_.height, size_.width, type_);
-    }
-
-    inline GpuMat::GpuMat(int rows_, int cols_, int type_, Scalar s_)
-        : flags(0), rows(0), cols(0), step(0), data(0), refcount(0), datastart(0), dataend(0)
-    {
-        if (rows_ > 0 && cols_ > 0)
-        {
-            create(rows_, cols_, type_);
-            setTo(s_);
-        }
-    }
-
-    inline GpuMat::GpuMat(Size size_, int type_, Scalar s_)
-        : flags(0), rows(0), cols(0), step(0), data(0), refcount(0), datastart(0), dataend(0)
-    {
-        if (size_.height > 0 && size_.width > 0)
-        {
-            create(size_.height, size_.width, type_);
-            setTo(s_);
-        }
-    }
-
-    inline GpuMat::~GpuMat()
-    {
-        release();
-    }
-
-    inline GpuMat GpuMat::clone() const
-    {
-        GpuMat m;
-        copyTo(m);
-        return m;
-    }
-
-    inline void GpuMat::assignTo(GpuMat& m, int _type) const
-    {
-        if (_type < 0)
-            m = *this;
-        else
-            convertTo(m, _type);
-    }
-
-    inline size_t GpuMat::step1() const
-    {
-        return step / elemSize1();
-    }
-
-    inline bool GpuMat::empty() const
-    {
-        return data == 0;
-    }
-
-    template<typename _Tp> inline _Tp* GpuMat::ptr(int y)
-    {
-        return (_Tp*)ptr(y);
-    }
-
-    template<typename _Tp> inline const _Tp* GpuMat::ptr(int y) const
-    {
-        return (const _Tp*)ptr(y);
-    }
-
-    inline void swap(GpuMat& a, GpuMat& b)
-    {
-        a.swap(b);
-    }
-
-    inline GpuMat GpuMat::row(int y) const
-    {
-        return GpuMat(*this, Range(y, y+1), Range::all());
-    }
-
-    inline GpuMat GpuMat::col(int x) const
-    {
-        return GpuMat(*this, Range::all(), Range(x, x+1));
-    }
-
-    inline GpuMat GpuMat::rowRange(int startrow, int endrow) const
-    {
-        return GpuMat(*this, Range(startrow, endrow), Range::all());
-    }
-
-    inline GpuMat GpuMat::rowRange(Range r) const
-    {
-        return GpuMat(*this, r, Range::all());
-    }
-
-    inline GpuMat GpuMat::colRange(int startcol, int endcol) const
-    {
-        return GpuMat(*this, Range::all(), Range(startcol, endcol));
-    }
-
-    inline GpuMat GpuMat::colRange(Range r) const
-    {
-        return GpuMat(*this, Range::all(), r);
-    }
-
-    inline void GpuMat::create(Size size_, int type_)
-    {
-        create(size_.height, size_.width, type_);
-    }
-
-    inline GpuMat GpuMat::operator()(Range _rowRange, Range _colRange) const
-    {
-        return GpuMat(*this, _rowRange, _colRange);
-    }
-
-    inline GpuMat GpuMat::operator()(Rect roi) const
-    {
-        return GpuMat(*this, roi);
-    }
-
-    inline bool GpuMat::isContinuous() const
-    {
-        return (flags & Mat::CONTINUOUS_FLAG) != 0;
-    }
-
-    inline size_t GpuMat::elemSize() const
-    {
-        return CV_ELEM_SIZE(flags);
-    }
-
-    inline size_t GpuMat::elemSize1() const
-    {
-        return CV_ELEM_SIZE1(flags);
-    }
-
-    inline int GpuMat::type() const
-    {
-        return CV_MAT_TYPE(flags);
-    }
-
-    inline int GpuMat::depth() const
-    {
-        return CV_MAT_DEPTH(flags);
-    }
-
-    inline int GpuMat::channels() const
-    {
-        return CV_MAT_CN(flags);
-    }
-
-    inline Size GpuMat::size() const
-    {
-        return Size(cols, rows);
-    }
-
-    inline uchar* GpuMat::ptr(int y)
-    {
-        CV_DbgAssert((unsigned)y < (unsigned)rows);
-        return data + step * y;
-    }
-
-    inline const uchar* GpuMat::ptr(int y) const
-    {
-        CV_DbgAssert((unsigned)y < (unsigned)rows);
-        return data + step * y;
-    }
-
-    inline GpuMat& GpuMat::operator = (Scalar s)
-    {
-        setTo(s);
-        return *this;
-    }
-
-    template <class T> inline GpuMat::operator PtrStepSz<T>() const
-    {
-        return PtrStepSz<T>(rows, cols, (T*)data, step);
-    }
-
-    template <class T> inline GpuMat::operator PtrStep<T>() const
-    {
-        return PtrStep<T>((T*)data, step);
-    }
-
-    template <class T> inline GpuMat::operator DevMem2D_<T>() const
-    {
-        return DevMem2D_<T>(rows, cols, (T*)data, step);
-    }
-
-    template <class T> inline GpuMat::operator PtrStep_<T>() const
-    {
-        return PtrStep_<T>(static_cast< DevMem2D_<T> >(*this));
-    }
-
-    inline GpuMat createContinuous(int rows, int cols, int type)
-    {
-        GpuMat m;
-        createContinuous(rows, cols, type, m);
-        return m;
-    }
-
-    inline void createContinuous(Size size, int type, GpuMat& m)
-    {
-        createContinuous(size.height, size.width, type, m);
-    }
-
-    inline GpuMat createContinuous(Size size, int type)
-    {
-        GpuMat m;
-        createContinuous(size, type, m);
-        return m;
-    }
-
-    inline void ensureSizeIsEnough(Size size, int type, GpuMat& m)
-    {
-        ensureSizeIsEnough(size.height, size.width, type, m);
-    }
-}}
-
-#endif // __cplusplus
-
-#endif // __OPENCV_GPUMAT_HPP__

include/opencv2/core/hal/hal.hpp

@@ -0,0 +1,250 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                          License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
+// Copyright (C) 2015, Itseez Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef OPENCV_HAL_HPP
+#define OPENCV_HAL_HPP
+
+#include "opencv2/core/cvdef.h"
+#include "opencv2/core/cvstd.hpp"
+#include "opencv2/core/hal/interface.h"
+
+namespace cv { namespace hal {
+
+//! @addtogroup core_hal_functions
+//! @{
+
+CV_EXPORTS int normHamming(const uchar* a, int n);
+CV_EXPORTS int normHamming(const uchar* a, const uchar* b, int n);
+
+CV_EXPORTS int normHamming(const uchar* a, int n, int cellSize);
+CV_EXPORTS int normHamming(const uchar* a, const uchar* b, int n, int cellSize);
+
+CV_EXPORTS int LU32f(float* A, size_t astep, int m, float* b, size_t bstep, int n);
+CV_EXPORTS int LU64f(double* A, size_t astep, int m, double* b, size_t bstep, int n);
+CV_EXPORTS bool Cholesky32f(float* A, size_t astep, int m, float* b, size_t bstep, int n);
+CV_EXPORTS bool Cholesky64f(double* A, size_t astep, int m, double* b, size_t bstep, int n);
+CV_EXPORTS void SVD32f(float* At, size_t astep, float* W, float* U, size_t ustep, float* Vt, size_t vstep, int m, int n, int flags);
+CV_EXPORTS void SVD64f(double* At, size_t astep, double* W, double* U, size_t ustep, double* Vt, size_t vstep, int m, int n, int flags);
+CV_EXPORTS int QR32f(float* A, size_t astep, int m, int n, int k, float* b, size_t bstep, float* hFactors);
+CV_EXPORTS int QR64f(double* A, size_t astep, int m, int n, int k, double* b, size_t bstep, double* hFactors);
+
+CV_EXPORTS void gemm32f(const float* src1, size_t src1_step, const float* src2, size_t src2_step,
+                        float alpha, const float* src3, size_t src3_step, float beta, float* dst, size_t dst_step,
+                        int m_a, int n_a, int n_d, int flags);
+CV_EXPORTS void gemm64f(const double* src1, size_t src1_step, const double* src2, size_t src2_step,
+                        double alpha, const double* src3, size_t src3_step, double beta, double* dst, size_t dst_step,
+                        int m_a, int n_a, int n_d, int flags);
+CV_EXPORTS void gemm32fc(const float* src1, size_t src1_step, const float* src2, size_t src2_step,
+                        float alpha, const float* src3, size_t src3_step, float beta, float* dst, size_t dst_step,
+                        int m_a, int n_a, int n_d, int flags);
+CV_EXPORTS void gemm64fc(const double* src1, size_t src1_step, const double* src2, size_t src2_step,
+                        double alpha, const double* src3, size_t src3_step, double beta, double* dst, size_t dst_step,
+                        int m_a, int n_a, int n_d, int flags);
+
+CV_EXPORTS int normL1_(const uchar* a, const uchar* b, int n);
+CV_EXPORTS float normL1_(const float* a, const float* b, int n);
+CV_EXPORTS float normL2Sqr_(const float* a, const float* b, int n);
+
+CV_EXPORTS void exp32f(const float* src, float* dst, int n);
+CV_EXPORTS void exp64f(const double* src, double* dst, int n);
+CV_EXPORTS void log32f(const float* src, float* dst, int n);
+CV_EXPORTS void log64f(const double* src, double* dst, int n);
+
+CV_EXPORTS void fastAtan32f(const float* y, const float* x, float* dst, int n, bool angleInDegrees);
+CV_EXPORTS void fastAtan64f(const double* y, const double* x, double* dst, int n, bool angleInDegrees);
+CV_EXPORTS void magnitude32f(const float* x, const float* y, float* dst, int n);
+CV_EXPORTS void magnitude64f(const double* x, const double* y, double* dst, int n);
+CV_EXPORTS void sqrt32f(const float* src, float* dst, int len);
+CV_EXPORTS void sqrt64f(const double* src, double* dst, int len);
+CV_EXPORTS void invSqrt32f(const float* src, float* dst, int len);
+CV_EXPORTS void invSqrt64f(const double* src, double* dst, int len);
+
+CV_EXPORTS void split8u(const uchar* src, uchar** dst, int len, int cn );
+CV_EXPORTS void split16u(const ushort* src, ushort** dst, int len, int cn );
+CV_EXPORTS void split32s(const int* src, int** dst, int len, int cn );
+CV_EXPORTS void split64s(const int64* src, int64** dst, int len, int cn );
+
+CV_EXPORTS void merge8u(const uchar** src, uchar* dst, int len, int cn );
+CV_EXPORTS void merge16u(const ushort** src, ushort* dst, int len, int cn );
+CV_EXPORTS void merge32s(const int** src, int* dst, int len, int cn );
+CV_EXPORTS void merge64s(const int64** src, int64* dst, int len, int cn );
+
+CV_EXPORTS void add8u( const uchar* src1, size_t step1, const uchar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void add8s( const schar* src1, size_t step1, const schar* src2, size_t step2, schar* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void add16u( const ushort* src1, size_t step1, const ushort* src2, size_t step2, ushort* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void add16s( const short* src1, size_t step1, const short* src2, size_t step2, short* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void add32s( const int* src1, size_t step1, const int* src2, size_t step2, int* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void add32f( const float* src1, size_t step1, const float* src2, size_t step2, float* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void add64f( const double* src1, size_t step1, const double* src2, size_t step2, double* dst, size_t step, int width, int height, void* );
+
+CV_EXPORTS void sub8u( const uchar* src1, size_t step1, const uchar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void sub8s( const schar* src1, size_t step1, const schar* src2, size_t step2, schar* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void sub16u( const ushort* src1, size_t step1, const ushort* src2, size_t step2, ushort* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void sub16s( const short* src1, size_t step1, const short* src2, size_t step2, short* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void sub32s( const int* src1, size_t step1, const int* src2, size_t step2, int* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void sub32f( const float* src1, size_t step1, const float* src2, size_t step2, float* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void sub64f( const double* src1, size_t step1, const double* src2, size_t step2, double* dst, size_t step, int width, int height, void* );
+
+CV_EXPORTS void max8u( const uchar* src1, size_t step1, const uchar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void max8s( const schar* src1, size_t step1, const schar* src2, size_t step2, schar* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void max16u( const ushort* src1, size_t step1, const ushort* src2, size_t step2, ushort* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void max16s( const short* src1, size_t step1, const short* src2, size_t step2, short* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void max32s( const int* src1, size_t step1, const int* src2, size_t step2, int* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void max32f( const float* src1, size_t step1, const float* src2, size_t step2, float* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void max64f( const double* src1, size_t step1, const double* src2, size_t step2, double* dst, size_t step, int width, int height, void* );
+
+CV_EXPORTS void min8u( const uchar* src1, size_t step1, const uchar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void min8s( const schar* src1, size_t step1, const schar* src2, size_t step2, schar* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void min16u( const ushort* src1, size_t step1, const ushort* src2, size_t step2, ushort* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void min16s( const short* src1, size_t step1, const short* src2, size_t step2, short* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void min32s( const int* src1, size_t step1, const int* src2, size_t step2, int* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void min32f( const float* src1, size_t step1, const float* src2, size_t step2, float* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void min64f( const double* src1, size_t step1, const double* src2, size_t step2, double* dst, size_t step, int width, int height, void* );
+
+CV_EXPORTS void absdiff8u( const uchar* src1, size_t step1, const uchar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void absdiff8s( const schar* src1, size_t step1, const schar* src2, size_t step2, schar* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void absdiff16u( const ushort* src1, size_t step1, const ushort* src2, size_t step2, ushort* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void absdiff16s( const short* src1, size_t step1, const short* src2, size_t step2, short* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void absdiff32s( const int* src1, size_t step1, const int* src2, size_t step2, int* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void absdiff32f( const float* src1, size_t step1, const float* src2, size_t step2, float* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void absdiff64f( const double* src1, size_t step1, const double* src2, size_t step2, double* dst, size_t step, int width, int height, void* );
+
+CV_EXPORTS void and8u( const uchar* src1, size_t step1, const uchar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void or8u( const uchar* src1, size_t step1, const uchar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void xor8u( const uchar* src1, size_t step1, const uchar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* );
+CV_EXPORTS void not8u( const uchar* src1, size_t step1, const uchar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* );
+
+CV_EXPORTS void cmp8u(const uchar* src1, size_t step1, const uchar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* _cmpop);
+CV_EXPORTS void cmp8s(const schar* src1, size_t step1, const schar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* _cmpop);
+CV_EXPORTS void cmp16u(const ushort* src1, size_t step1, const ushort* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* _cmpop);
+CV_EXPORTS void cmp16s(const short* src1, size_t step1, const short* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* _cmpop);
+CV_EXPORTS void cmp32s(const int* src1, size_t step1, const int* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* _cmpop);
+CV_EXPORTS void cmp32f(const float* src1, size_t step1, const float* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* _cmpop);
+CV_EXPORTS void cmp64f(const double* src1, size_t step1, const double* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* _cmpop);
+
+CV_EXPORTS void mul8u( const uchar* src1, size_t step1, const uchar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* scale);
+CV_EXPORTS void mul8s( const schar* src1, size_t step1, const schar* src2, size_t step2, schar* dst, size_t step, int width, int height, void* scale);
+CV_EXPORTS void mul16u( const ushort* src1, size_t step1, const ushort* src2, size_t step2, ushort* dst, size_t step, int width, int height, void* scale);
+CV_EXPORTS void mul16s( const short* src1, size_t step1, const short* src2, size_t step2, short* dst, size_t step, int width, int height, void* scale);
+CV_EXPORTS void mul32s( const int* src1, size_t step1, const int* src2, size_t step2, int* dst, size_t step, int width, int height, void* scale);
+CV_EXPORTS void mul32f( const float* src1, size_t step1, const float* src2, size_t step2, float* dst, size_t step, int width, int height, void* scale);
+CV_EXPORTS void mul64f( const double* src1, size_t step1, const double* src2, size_t step2, double* dst, size_t step, int width, int height, void* scale);
+
+CV_EXPORTS void div8u( const uchar* src1, size_t step1, const uchar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* scale);
+CV_EXPORTS void div8s( const schar* src1, size_t step1, const schar* src2, size_t step2, schar* dst, size_t step, int width, int height, void* scale);
+CV_EXPORTS void div16u( const ushort* src1, size_t step1, const ushort* src2, size_t step2, ushort* dst, size_t step, int width, int height, void* scale);
+CV_EXPORTS void div16s( const short* src1, size_t step1, const short* src2, size_t step2, short* dst, size_t step, int width, int height, void* scale);
+CV_EXPORTS void div32s( const int* src1, size_t step1, const int* src2, size_t step2, int* dst, size_t step, int width, int height, void* scale);
+CV_EXPORTS void div32f( const float* src1, size_t step1, const float* src2, size_t step2, float* dst, size_t step, int width, int height, void* scale);
+CV_EXPORTS void div64f( const double* src1, size_t step1, const double* src2, size_t step2, double* dst, size_t step, int width, int height, void* scale);
+
+CV_EXPORTS void recip8u( const uchar *, size_t, const uchar * src2, size_t step2, uchar* dst, size_t step, int width, int height, void* scale);
+CV_EXPORTS void recip8s( const schar *, size_t, const schar * src2, size_t step2, schar* dst, size_t step, int width, int height, void* scale);
+CV_EXPORTS void recip16u( const ushort *, size_t, const ushort * src2, size_t step2, ushort* dst, size_t step, int width, int height, void* scale);
+CV_EXPORTS void recip16s( const short *, size_t, const short * src2, size_t step2, short* dst, size_t step, int width, int height, void* scale);
+CV_EXPORTS void recip32s( const int *, size_t, const int * src2, size_t step2, int* dst, size_t step, int width, int height, void* scale);
+CV_EXPORTS void recip32f( const float *, size_t, const float * src2, size_t step2, float* dst, size_t step, int width, int height, void* scale);
+CV_EXPORTS void recip64f( const double *, size_t, const double * src2, size_t step2, double* dst, size_t step, int width, int height, void* scale);
+
+CV_EXPORTS void addWeighted8u( const uchar* src1, size_t step1, const uchar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* _scalars );
+CV_EXPORTS void addWeighted8s( const schar* src1, size_t step1, const schar* src2, size_t step2, schar* dst, size_t step, int width, int height, void* scalars );
+CV_EXPORTS void addWeighted16u( const ushort* src1, size_t step1, const ushort* src2, size_t step2, ushort* dst, size_t step, int width, int height, void* scalars );
+CV_EXPORTS void addWeighted16s( const short* src1, size_t step1, const short* src2, size_t step2, short* dst, size_t step, int width, int height, void* scalars );
+CV_EXPORTS void addWeighted32s( const int* src1, size_t step1, const int* src2, size_t step2, int* dst, size_t step, int width, int height, void* scalars );
+CV_EXPORTS void addWeighted32f( const float* src1, size_t step1, const float* src2, size_t step2, float* dst, size_t step, int width, int height, void* scalars );
+CV_EXPORTS void addWeighted64f( const double* src1, size_t step1, const double* src2, size_t step2, double* dst, size_t step, int width, int height, void* scalars );
+
+struct CV_EXPORTS DFT1D
+{
+    static Ptr<DFT1D> create(int len, int count, int depth, int flags, bool * useBuffer = 0);
+    virtual void apply(const uchar *src, uchar *dst) = 0;
+    virtual ~DFT1D() {}
+};
+
+struct CV_EXPORTS DFT2D
+{
+    static Ptr<DFT2D> create(int width, int height, int depth,
+                             int src_channels, int dst_channels,
+                             int flags, int nonzero_rows = 0);
+    virtual void apply(const uchar *src_data, size_t src_step, uchar *dst_data, size_t dst_step) = 0;
+    virtual ~DFT2D() {}
+};
+
+struct CV_EXPORTS DCT2D
+{
+    static Ptr<DCT2D> create(int width, int height, int depth, int flags);
+    virtual void apply(const uchar *src_data, size_t src_step, uchar *dst_data, size_t dst_step) = 0;
+    virtual ~DCT2D() {}
+};
+
+//! @} core_hal
+
+//=============================================================================
+// for binary compatibility with 3.0
+
+//! @cond IGNORED
+
+CV_EXPORTS int LU(float* A, size_t astep, int m, float* b, size_t bstep, int n);
+CV_EXPORTS int LU(double* A, size_t astep, int m, double* b, size_t bstep, int n);
+CV_EXPORTS bool Cholesky(float* A, size_t astep, int m, float* b, size_t bstep, int n);
+CV_EXPORTS bool Cholesky(double* A, size_t astep, int m, double* b, size_t bstep, int n);
+
+CV_EXPORTS void exp(const float* src, float* dst, int n);
+CV_EXPORTS void exp(const double* src, double* dst, int n);
+CV_EXPORTS void log(const float* src, float* dst, int n);
+CV_EXPORTS void log(const double* src, double* dst, int n);
+
+CV_EXPORTS void fastAtan2(const float* y, const float* x, float* dst, int n, bool angleInDegrees);
+CV_EXPORTS void magnitude(const float* x, const float* y, float* dst, int n);
+CV_EXPORTS void magnitude(const double* x, const double* y, double* dst, int n);
+CV_EXPORTS void sqrt(const float* src, float* dst, int len);
+CV_EXPORTS void sqrt(const double* src, double* dst, int len);
+CV_EXPORTS void invSqrt(const float* src, float* dst, int len);
+CV_EXPORTS void invSqrt(const double* src, double* dst, int len);
+
+//! @endcond
+
+}} //cv::hal
+
+#endif //OPENCV_HAL_HPP

include/opencv2/core/hal/interface.h

@@ -0,0 +1,178 @@
+#ifndef OPENCV_CORE_HAL_INTERFACE_H
+#define OPENCV_CORE_HAL_INTERFACE_H
+
+//! @addtogroup core_hal_interface
+//! @{
+
+//! @name Return codes
+//! @{
+#define CV_HAL_ERROR_OK 0
+#define CV_HAL_ERROR_NOT_IMPLEMENTED 1
+#define CV_HAL_ERROR_UNKNOWN -1
+//! @}
+
+#ifdef __cplusplus
+#include <cstddef>
+#else
+#include <stddef.h>
+#include <stdbool.h>
+#endif
+
+//! @name Data types
+//! primitive types
+//! - schar  - signed 1 byte integer
+//! - uchar  - unsigned 1 byte integer
+//! - short  - signed 2 byte integer
+//! - ushort - unsigned 2 byte integer
+//! - int    - signed 4 byte integer
+//! - uint   - unsigned 4 byte integer
+//! - int64  - signed 8 byte integer
+//! - uint64 - unsigned 8 byte integer
+//! @{
+#if !defined _MSC_VER && !defined __BORLANDC__
+#  if defined __cplusplus && __cplusplus >= 201103L && !defined __APPLE__
+#    include <cstdint>
+     typedef std::uint32_t uint;
+#  else
+#    include <stdint.h>
+     typedef uint32_t uint;
+#  endif
+#else
+   typedef unsigned uint;
+#endif
+
+typedef signed char schar;
+
+#ifndef __IPL_H__
+   typedef unsigned char uchar;
+   typedef unsigned short ushort;
+#endif
+
+#if defined _MSC_VER || defined __BORLANDC__
+   typedef __int64 int64;
+   typedef unsigned __int64 uint64;
+#  define CV_BIG_INT(n)   n##I64
+#  define CV_BIG_UINT(n)  n##UI64
+#else
+   typedef int64_t int64;
+   typedef uint64_t uint64;
+#  define CV_BIG_INT(n)   n##LL
+#  define CV_BIG_UINT(n)  n##ULL
+#endif
+
+#define CV_CN_MAX     512
+#define CV_CN_SHIFT   3
+#define CV_DEPTH_MAX  (1 << CV_CN_SHIFT)
+
+#define CV_8U   0
+#define CV_8S   1
+#define CV_16U  2
+#define CV_16S  3
+#define CV_32S  4
+#define CV_32F  5
+#define CV_64F  6
+#define CV_USRTYPE1 7
+
+#define CV_MAT_DEPTH_MASK       (CV_DEPTH_MAX - 1)
+#define CV_MAT_DEPTH(flags)     ((flags) & CV_MAT_DEPTH_MASK)
+
+#define CV_MAKETYPE(depth,cn) (CV_MAT_DEPTH(depth) + (((cn)-1) << CV_CN_SHIFT))
+#define CV_MAKE_TYPE CV_MAKETYPE
+
+#define CV_8UC1 CV_MAKETYPE(CV_8U,1)
+#define CV_8UC2 CV_MAKETYPE(CV_8U,2)
+#define CV_8UC3 CV_MAKETYPE(CV_8U,3)
+#define CV_8UC4 CV_MAKETYPE(CV_8U,4)
+#define CV_8UC(n) CV_MAKETYPE(CV_8U,(n))
+
+#define CV_8SC1 CV_MAKETYPE(CV_8S,1)
+#define CV_8SC2 CV_MAKETYPE(CV_8S,2)
+#define CV_8SC3 CV_MAKETYPE(CV_8S,3)
+#define CV_8SC4 CV_MAKETYPE(CV_8S,4)
+#define CV_8SC(n) CV_MAKETYPE(CV_8S,(n))
+
+#define CV_16UC1 CV_MAKETYPE(CV_16U,1)
+#define CV_16UC2 CV_MAKETYPE(CV_16U,2)
+#define CV_16UC3 CV_MAKETYPE(CV_16U,3)
+#define CV_16UC4 CV_MAKETYPE(CV_16U,4)
+#define CV_16UC(n) CV_MAKETYPE(CV_16U,(n))
+
+#define CV_16SC1 CV_MAKETYPE(CV_16S,1)
+#define CV_16SC2 CV_MAKETYPE(CV_16S,2)
+#define CV_16SC3 CV_MAKETYPE(CV_16S,3)
+#define CV_16SC4 CV_MAKETYPE(CV_16S,4)
+#define CV_16SC(n) CV_MAKETYPE(CV_16S,(n))
+
+#define CV_32SC1 CV_MAKETYPE(CV_32S,1)
+#define CV_32SC2 CV_MAKETYPE(CV_32S,2)
+#define CV_32SC3 CV_MAKETYPE(CV_32S,3)
+#define CV_32SC4 CV_MAKETYPE(CV_32S,4)
+#define CV_32SC(n) CV_MAKETYPE(CV_32S,(n))
+
+#define CV_32FC1 CV_MAKETYPE(CV_32F,1)
+#define CV_32FC2 CV_MAKETYPE(CV_32F,2)
+#define CV_32FC3 CV_MAKETYPE(CV_32F,3)
+#define CV_32FC4 CV_MAKETYPE(CV_32F,4)
+#define CV_32FC(n) CV_MAKETYPE(CV_32F,(n))
+
+#define CV_64FC1 CV_MAKETYPE(CV_64F,1)
+#define CV_64FC2 CV_MAKETYPE(CV_64F,2)
+#define CV_64FC3 CV_MAKETYPE(CV_64F,3)
+#define CV_64FC4 CV_MAKETYPE(CV_64F,4)
+#define CV_64FC(n) CV_MAKETYPE(CV_64F,(n))
+//! @}
+
+//! @name Comparison operation
+//! @sa cv::CmpTypes
+//! @{
+#define CV_HAL_CMP_EQ 0
+#define CV_HAL_CMP_GT 1
+#define CV_HAL_CMP_GE 2
+#define CV_HAL_CMP_LT 3
+#define CV_HAL_CMP_LE 4
+#define CV_HAL_CMP_NE 5
+//! @}
+
+//! @name Border processing modes
+//! @sa cv::BorderTypes
+//! @{
+#define CV_HAL_BORDER_CONSTANT 0
+#define CV_HAL_BORDER_REPLICATE 1
+#define CV_HAL_BORDER_REFLECT 2
+#define CV_HAL_BORDER_WRAP 3
+#define CV_HAL_BORDER_REFLECT_101 4
+#define CV_HAL_BORDER_TRANSPARENT 5
+#define CV_HAL_BORDER_ISOLATED 16
+//! @}
+
+//! @name DFT flags
+//! @{
+#define CV_HAL_DFT_INVERSE        1
+#define CV_HAL_DFT_SCALE          2
+#define CV_HAL_DFT_ROWS           4
+#define CV_HAL_DFT_COMPLEX_OUTPUT 16
+#define CV_HAL_DFT_REAL_OUTPUT    32
+#define CV_HAL_DFT_TWO_STAGE      64
+#define CV_HAL_DFT_STAGE_COLS    128
+#define CV_HAL_DFT_IS_CONTINUOUS 512
+#define CV_HAL_DFT_IS_INPLACE 1024
+//! @}
+
+//! @name SVD flags
+//! @{
+#define CV_HAL_SVD_NO_UV    1
+#define CV_HAL_SVD_SHORT_UV 2
+#define CV_HAL_SVD_MODIFY_A 4
+#define CV_HAL_SVD_FULL_UV  8
+//! @}
+
+//! @name Gemm flags
+//! @{
+#define CV_HAL_GEMM_1_T 1
+#define CV_HAL_GEMM_2_T 2
+#define CV_HAL_GEMM_3_T 4
+//! @}
+
+//! @}
+
+#endif

include/opencv2/core/hal/intrin.hpp

@@ -0,0 +1,414 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                          License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
+// Copyright (C) 2015, Itseez Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef OPENCV_HAL_INTRIN_HPP
+#define OPENCV_HAL_INTRIN_HPP
+
+#include <cmath>
+#include <float.h>
+#include <stdlib.h>
+#include "opencv2/core/cvdef.h"
+
+#define OPENCV_HAL_ADD(a, b) ((a) + (b))
+#define OPENCV_HAL_AND(a, b) ((a) & (b))
+#define OPENCV_HAL_NOP(a) (a)
+#define OPENCV_HAL_1ST(a, b) (a)
+
+// unlike HAL API, which is in cv::hal,
+// we put intrinsics into cv namespace to make its
+// access from within opencv code more accessible
+namespace cv {
+
+//! @addtogroup core_hal_intrin
+//! @{
+
+//! @cond IGNORED
+template<typename _Tp> struct V_TypeTraits
+{
+    typedef _Tp int_type;
+    typedef _Tp uint_type;
+    typedef _Tp abs_type;
+    typedef _Tp sum_type;
+
+    enum { delta = 0, shift = 0 };
+
+    static int_type reinterpret_int(_Tp x) { return x; }
+    static uint_type reinterpet_uint(_Tp x) { return x; }
+    static _Tp reinterpret_from_int(int_type x) { return (_Tp)x; }
+};
+
+template<> struct V_TypeTraits<uchar>
+{
+    typedef uchar value_type;
+    typedef schar int_type;
+    typedef uchar uint_type;
+    typedef uchar abs_type;
+    typedef int sum_type;
+
+    typedef ushort w_type;
+    typedef unsigned q_type;
+
+    enum { delta = 128, shift = 8 };
+
+    static int_type reinterpret_int(value_type x) { return (int_type)x; }
+    static uint_type reinterpret_uint(value_type x) { return (uint_type)x; }
+    static value_type reinterpret_from_int(int_type x) { return (value_type)x; }
+};
+
+template<> struct V_TypeTraits<schar>
+{
+    typedef schar value_type;
+    typedef schar int_type;
+    typedef uchar uint_type;
+    typedef uchar abs_type;
+    typedef int sum_type;
+
+    typedef short w_type;
+    typedef int q_type;
+
+    enum { delta = 128, shift = 8 };
+
+    static int_type reinterpret_int(value_type x) { return (int_type)x; }
+    static uint_type reinterpret_uint(value_type x) { return (uint_type)x; }
+    static value_type reinterpret_from_int(int_type x) { return (value_type)x; }
+};
+
+template<> struct V_TypeTraits<ushort>
+{
+    typedef ushort value_type;
+    typedef short int_type;
+    typedef ushort uint_type;
+    typedef ushort abs_type;
+    typedef int sum_type;
+
+    typedef unsigned w_type;
+    typedef uchar nu_type;
+
+    enum { delta = 32768, shift = 16 };
+
+    static int_type reinterpret_int(value_type x) { return (int_type)x; }
+    static uint_type reinterpret_uint(value_type x) { return (uint_type)x; }
+    static value_type reinterpret_from_int(int_type x) { return (value_type)x; }
+};
+
+template<> struct V_TypeTraits<short>
+{
+    typedef short value_type;
+    typedef short int_type;
+    typedef ushort uint_type;
+    typedef ushort abs_type;
+    typedef int sum_type;
+
+    typedef int w_type;
+    typedef uchar nu_type;
+    typedef schar n_type;
+
+    enum { delta = 128, shift = 8 };
+
+    static int_type reinterpret_int(value_type x) { return (int_type)x; }
+    static uint_type reinterpret_uint(value_type x) { return (uint_type)x; }
+    static value_type reinterpret_from_int(int_type x) { return (value_type)x; }
+};
+
+template<> struct V_TypeTraits<unsigned>
+{
+    typedef unsigned value_type;
+    typedef int int_type;
+    typedef unsigned uint_type;
+    typedef unsigned abs_type;
+    typedef unsigned sum_type;
+
+    typedef uint64 w_type;
+    typedef ushort nu_type;
+
+    static int_type reinterpret_int(value_type x) { return (int_type)x; }
+    static uint_type reinterpret_uint(value_type x) { return (uint_type)x; }
+    static value_type reinterpret_from_int(int_type x) { return (value_type)x; }
+};
+
+template<> struct V_TypeTraits<int>
+{
+    typedef int value_type;
+    typedef int int_type;
+    typedef unsigned uint_type;
+    typedef unsigned abs_type;
+    typedef int sum_type;
+
+    typedef int64 w_type;
+    typedef short n_type;
+    typedef ushort nu_type;
+
+    static int_type reinterpret_int(value_type x) { return (int_type)x; }
+    static uint_type reinterpret_uint(value_type x) { return (uint_type)x; }
+    static value_type reinterpret_from_int(int_type x) { return (value_type)x; }
+};
+
+template<> struct V_TypeTraits<uint64>
+{
+    typedef uint64 value_type;
+    typedef int64 int_type;
+    typedef uint64 uint_type;
+    typedef uint64 abs_type;
+    typedef uint64 sum_type;
+
+    typedef unsigned nu_type;
+
+    static int_type reinterpret_int(value_type x) { return (int_type)x; }
+    static uint_type reinterpret_uint(value_type x) { return (uint_type)x; }
+    static value_type reinterpret_from_int(int_type x) { return (value_type)x; }
+};
+
+template<> struct V_TypeTraits<int64>
+{
+    typedef int64 value_type;
+    typedef int64 int_type;
+    typedef uint64 uint_type;
+    typedef uint64 abs_type;
+    typedef int64 sum_type;
+
+    typedef int nu_type;
+
+    static int_type reinterpret_int(value_type x) { return (int_type)x; }
+    static uint_type reinterpret_uint(value_type x) { return (uint_type)x; }
+    static value_type reinterpret_from_int(int_type x) { return (value_type)x; }
+};
+
+
+template<> struct V_TypeTraits<float>
+{
+    typedef float value_type;
+    typedef int int_type;
+    typedef unsigned uint_type;
+    typedef float abs_type;
+    typedef float sum_type;
+
+    typedef double w_type;
+
+    static int_type reinterpret_int(value_type x)
+    {
+        Cv32suf u;
+        u.f = x;
+        return u.i;
+    }
+    static uint_type reinterpet_uint(value_type x)
+    {
+        Cv32suf u;
+        u.f = x;
+        return u.u;
+    }
+    static value_type reinterpret_from_int(int_type x)
+    {
+        Cv32suf u;
+        u.i = x;
+        return u.f;
+    }
+};
+
+template<> struct V_TypeTraits<double>
+{
+    typedef double value_type;
+    typedef int64 int_type;
+    typedef uint64 uint_type;
+    typedef double abs_type;
+    typedef double sum_type;
+    static int_type reinterpret_int(value_type x)
+    {
+        Cv64suf u;
+        u.f = x;
+        return u.i;
+    }
+    static uint_type reinterpet_uint(value_type x)
+    {
+        Cv64suf u;
+        u.f = x;
+        return u.u;
+    }
+    static value_type reinterpret_from_int(int_type x)
+    {
+        Cv64suf u;
+        u.i = x;
+        return u.f;
+    }
+};
+
+template <typename T> struct V_SIMD128Traits
+{
+    enum { nlanes = 16 / sizeof(T) };
+};
+
+//! @endcond
+
+//! @}
+
+}
+
+#ifdef CV_DOXYGEN
+#   undef CV_SSE2
+#   undef CV_NEON
+#endif
+
+#if CV_SSE2
+
+#include "opencv2/core/hal/intrin_sse.hpp"
+
+#elif CV_NEON
+
+#include "opencv2/core/hal/intrin_neon.hpp"
+
+#else
+
+#include "opencv2/core/hal/intrin_cpp.hpp"
+
+#endif
+
+//! @addtogroup core_hal_intrin
+//! @{
+
+#ifndef CV_SIMD128
+//! Set to 1 if current compiler supports vector extensions (NEON or SSE is enabled)
+#define CV_SIMD128 0
+#endif
+
+#ifndef CV_SIMD128_64F
+//! Set to 1 if current intrinsics implementation supports 64-bit float vectors
+#define CV_SIMD128_64F 0
+#endif
+
+//! @}
+
+//==================================================================================================
+
+//! @cond IGNORED
+
+namespace cv {
+
+template <typename R> struct V_RegTrait128;
+
+template <> struct V_RegTrait128<uchar> {
+    typedef v_uint8x16 reg;
+    typedef v_uint16x8 w_reg;
+    typedef v_uint32x4 q_reg;
+    typedef v_uint8x16 u_reg;
+    static v_uint8x16 zero() { return v_setzero_u8(); }
+    static v_uint8x16 all(uchar val) { return v_setall_u8(val); }
+};
+
+template <> struct V_RegTrait128<schar> {
+    typedef v_int8x16 reg;
+    typedef v_int16x8 w_reg;
+    typedef v_int32x4 q_reg;
+    typedef v_uint8x16 u_reg;
+    static v_int8x16 zero() { return v_setzero_s8(); }
+    static v_int8x16 all(schar val) { return v_setall_s8(val); }
+};
+
+template <> struct V_RegTrait128<ushort> {
+    typedef v_uint16x8 reg;
+    typedef v_uint32x4 w_reg;
+    typedef v_int16x8 int_reg;
+    typedef v_uint16x8 u_reg;
+    static v_uint16x8 zero() { return v_setzero_u16(); }
+    static v_uint16x8 all(ushort val) { return v_setall_u16(val); }
+};
+
+template <> struct V_RegTrait128<short> {
+    typedef v_int16x8 reg;
+    typedef v_int32x4 w_reg;
+    typedef v_uint16x8 u_reg;
+    static v_int16x8 zero() { return v_setzero_s16(); }
+    static v_int16x8 all(short val) { return v_setall_s16(val); }
+};
+
+template <> struct V_RegTrait128<unsigned> {
+    typedef v_uint32x4 reg;
+    typedef v_uint64x2 w_reg;
+    typedef v_int32x4 int_reg;
+    typedef v_uint32x4 u_reg;
+    static v_uint32x4 zero() { return v_setzero_u32(); }
+    static v_uint32x4 all(unsigned val) { return v_setall_u32(val); }
+};
+
+template <> struct V_RegTrait128<int> {
+    typedef v_int32x4 reg;
+    typedef v_int64x2 w_reg;
+    typedef v_uint32x4 u_reg;
+    static v_int32x4 zero() { return v_setzero_s32(); }
+    static v_int32x4 all(int val) { return v_setall_s32(val); }
+};
+
+template <> struct V_RegTrait128<uint64> {
+    typedef v_uint64x2 reg;
+    static v_uint64x2 zero() { return v_setzero_u64(); }
+    static v_uint64x2 all(uint64 val) { return v_setall_u64(val); }
+};
+
+template <> struct V_RegTrait128<int64> {
+    typedef v_int64x2 reg;
+    static v_int64x2 zero() { return v_setzero_s64(); }
+    static v_int64x2 all(int64 val) { return v_setall_s64(val); }
+};
+
+template <> struct V_RegTrait128<float> {
+    typedef v_float32x4 reg;
+    typedef v_int32x4 int_reg;
+    typedef v_float32x4 u_reg;
+    static v_float32x4 zero() { return v_setzero_f32(); }
+    static v_float32x4 all(float val) { return v_setall_f32(val); }
+};
+
+#if CV_SIMD128_64F
+template <> struct V_RegTrait128<double> {
+    typedef v_float64x2 reg;
+    typedef v_int32x4 int_reg;
+    typedef v_float64x2 u_reg;
+    static v_float64x2 zero() { return v_setzero_f64(); }
+    static v_float64x2 all(double val) { return v_setall_f64(val); }
+};
+#endif
+
+} // cv::
+
+//! @endcond
+
+#endif

include/opencv2/core/internal.hpp

@@ -1,781 +0,0 @@
-/*M///////////////////////////////////////////////////////////////////////////////////////
-//
-//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
-//
-//  By downloading, copying, installing or using the software you agree to this license.
-//  If you do not agree to this license, do not download, install,
-//  copy or use the software.
-//
-//
-//                           License Agreement
-//                For Open Source Computer Vision Library
-//
-// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
-// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
-// Third party copyrights are property of their respective owners.
-//
-// Redistribution and use in source and binary forms, with or without modification,
-// are permitted provided that the following conditions are met:
-//
-//   * Redistribution's of source code must retain the above copyright notice,
-//     this list of conditions and the following disclaimer.
-//
-//   * Redistribution's in binary form must reproduce the above copyright notice,
-//     this list of conditions and the following disclaimer in the documentation
-//     and/or other materials provided with the distribution.
-//
-//   * The name of the copyright holders may not be used to endorse or promote products
-//     derived from this software without specific prior written permission.
-//
-// This software is provided by the copyright holders and contributors "as is" and
-// any express or implied warranties, including, but not limited to, the implied
-// warranties of merchantability and fitness for a particular purpose are disclaimed.
-// In no event shall the Intel Corporation or contributors be liable for any direct,
-// indirect, incidental, special, exemplary, or consequential damages
-// (including, but not limited to, procurement of substitute goods or services;
-// loss of use, data, or profits; or business interruption) however caused
-// and on any theory of liability, whether in contract, strict liability,
-// or tort (including negligence or otherwise) arising in any way out of
-// the use of this software, even if advised of the possibility of such damage.
-//
-//M*/
-
-/* The header is for internal use and it is likely to change.
-   It contains some macro definitions that are used in cxcore, cv, cvaux
-   and, probably, other libraries. If you need some of this functionality,
-   the safe way is to copy it into your code and rename the macros.
-*/
-#ifndef __OPENCV_CORE_INTERNAL_HPP__
-#define __OPENCV_CORE_INTERNAL_HPP__
-
-#include <vector>
-
-#include "opencv2/core/core.hpp"
-#include "opencv2/core/types_c.h"
-
-#if defined WIN32 || defined _WIN32
-#  ifndef WIN32
-#    define WIN32
-#  endif
-#  ifndef _WIN32
-#    define _WIN32
-#  endif
-#endif
-
-#if !defined WIN32 && !defined WINCE
-#  include <pthread.h>
-#endif
-
-#ifdef __BORLANDC__
-#  ifndef WIN32
-#    define WIN32
-#  endif
-#  ifndef _WIN32
-#    define _WIN32
-#  endif
-#  define CV_DLL
-#  undef _CV_ALWAYS_PROFILE_
-#  define _CV_ALWAYS_NO_PROFILE_
-#endif
-
-#ifndef FALSE
-#  define FALSE 0
-#endif
-#ifndef TRUE
-#  define TRUE 1
-#endif
-
-#define __BEGIN__ __CV_BEGIN__
-#define __END__ __CV_END__
-#define EXIT __CV_EXIT__
-
-#ifdef HAVE_IPP
-#  include "ipp.h"
-
-CV_INLINE IppiSize ippiSize(int width, int height)
-{
-    IppiSize size = { width, height };
-    return size;
-}
-#endif
-
-#ifndef IPPI_CALL
-#  define IPPI_CALL(func) CV_Assert((func) >= 0)
-#endif
-
-#if defined __SSE2__ || defined _M_X64  || (defined _M_IX86_FP && _M_IX86_FP >= 2)
-#  include "emmintrin.h"
-#  define CV_SSE 1
-#  define CV_SSE2 1
-#  if defined __SSE3__ || (defined _MSC_VER && _MSC_VER >= 1500)
-#    include "pmmintrin.h"
-#    define CV_SSE3 1
-#  endif
-#  if defined __SSSE3__  || (defined _MSC_VER && _MSC_VER >= 1500)
-#    include "tmmintrin.h"
-#    define CV_SSSE3 1
-#  endif
-#  if defined __SSE4_1__ || (defined _MSC_VER && _MSC_VER >= 1500)
-#    include <smmintrin.h>
-#    define CV_SSE4_1 1
-#  endif
-#  if defined __SSE4_2__ || (defined _MSC_VER && _MSC_VER >= 1500)
-#    include <nmmintrin.h>
-#    define CV_SSE4_2 1
-#  endif
-#  if defined __AVX__ || (defined _MSC_FULL_VER && _MSC_FULL_VER >= 160040219)
-// MS Visual Studio 2010 (2012?) has no macro pre-defined to identify the use of /arch:AVX
-// See: http://connect.microsoft.com/VisualStudio/feedback/details/605858/arch-avx-should-define-a-predefined-macro-in-x64-and-set-a-unique-value-for-m-ix86-fp-in-win32
-#    include <immintrin.h>
-#    define CV_AVX 1
-#    if defined(_XCR_XFEATURE_ENABLED_MASK)
-#      define __xgetbv() _xgetbv(_XCR_XFEATURE_ENABLED_MASK)
-#    else
-#      define __xgetbv() 0
-#    endif
-#  endif
-#endif
-
-
-#if (defined WIN32 || defined _WIN32) && defined(_M_ARM)
-# include <Intrin.h>
-# include "arm_neon.h"
-# define CV_NEON 1
-# define CPU_HAS_NEON_FEATURE (true)
-#elif defined(__ARM_NEON__)
-#  include <arm_neon.h>
-#  define CV_NEON 1
-#  define CPU_HAS_NEON_FEATURE (true)
-#endif
-
-#ifndef CV_SSE
-#  define CV_SSE 0
-#endif
-#ifndef CV_SSE2
-#  define CV_SSE2 0
-#endif
-#ifndef CV_SSE3
-#  define CV_SSE3 0
-#endif
-#ifndef CV_SSSE3
-#  define CV_SSSE3 0
-#endif
-#ifndef CV_SSE4_1
-#  define CV_SSE4_1 0
-#endif
-#ifndef CV_SSE4_2
-#  define CV_SSE4_2 0
-#endif
-#ifndef CV_AVX
-#  define CV_AVX 0
-#endif
-#ifndef CV_NEON
-#  define CV_NEON 0
-#endif
-
-#ifdef HAVE_TBB
-#  include "tbb/tbb_stddef.h"
-#  if TBB_VERSION_MAJOR*100 + TBB_VERSION_MINOR >= 202
-#    include "tbb/tbb.h"
-#    include "tbb/task.h"
-#    undef min
-#    undef max
-#  else
-#    undef HAVE_TBB
-#  endif
-#endif
-
-#ifdef HAVE_EIGEN
-#  if defined __GNUC__ && defined __APPLE__
-#    pragma GCC diagnostic ignored "-Wshadow"
-#  endif
-#  include <Eigen/Core>
-#  include "opencv2/core/eigen.hpp"
-#endif
-
-#ifdef __cplusplus
-
-namespace cv
-{
-#ifdef HAVE_TBB
-
-    typedef tbb::blocked_range<int> BlockedRange;
-
-    template<typename Body> static inline
-    void parallel_for( const BlockedRange& range, const Body& body )
-    {
-        tbb::parallel_for(range, body);
-    }
-
-    template<typename Iterator, typename Body> static inline
-    void parallel_do( Iterator first, Iterator last, const Body& body )
-    {
-        tbb::parallel_do(first, last, body);
-    }
-
-    typedef tbb::split Split;
-
-    template<typename Body> static inline
-    void parallel_reduce( const BlockedRange& range, Body& body )
-    {
-        tbb::parallel_reduce(range, body);
-    }
-
-    typedef tbb::concurrent_vector<Rect> ConcurrentRectVector;
-    typedef tbb::concurrent_vector<double> ConcurrentDoubleVector;
-#else
-    class BlockedRange
-    {
-    public:
-        BlockedRange() : _begin(0), _end(0), _grainsize(0) {}
-        BlockedRange(int b, int e, int g=1) : _begin(b), _end(e), _grainsize(g) {}
-        int begin() const { return _begin; }
-        int end() const { return _end; }
-        int grainsize() const { return _grainsize; }
-
-    protected:
-        int _begin, _end, _grainsize;
-    };
-
-    template<typename Body> static inline
-    void parallel_for( const BlockedRange& range, const Body& body )
-    {
-        body(range);
-    }
-    typedef std::vector<Rect> ConcurrentRectVector;
-    typedef std::vector<double> ConcurrentDoubleVector;
-
-    template<typename Iterator, typename Body> static inline
-    void parallel_do( Iterator first, Iterator last, const Body& body )
-    {
-        for( ; first != last; ++first )
-            body(*first);
-    }
-
-    class Split {};
-
-    template<typename Body> static inline
-    void parallel_reduce( const BlockedRange& range, Body& body )
-    {
-        body(range);
-    }
-#endif
-
-    // Returns a static string if there is a parallel framework,
-    // NULL otherwise.
-    CV_EXPORTS const char* currentParallelFramework();
-} //namespace cv
-
-#define CV_INIT_ALGORITHM(classname, algname, memberinit) \
-    static ::cv::Algorithm* create##classname() \
-    { \
-        return new classname; \
-    } \
-    \
-    static ::cv::AlgorithmInfo& classname##_info() \
-    { \
-        static ::cv::AlgorithmInfo classname##_info_var(algname, create##classname); \
-        return classname##_info_var; \
-    } \
-    \
-    static ::cv::AlgorithmInfo& classname##_info_auto = classname##_info(); \
-    \
-    ::cv::AlgorithmInfo* classname::info() const \
-    { \
-        static volatile bool initialized = false; \
-        \
-        if( !initialized ) \
-        { \
-            initialized = true; \
-            classname obj; \
-            memberinit; \
-        } \
-        return &classname##_info(); \
-    }
-
-#endif //__cplusplus
-
-/* maximal size of vector to run matrix operations on it inline (i.e. w/o ipp calls) */
-#define  CV_MAX_INLINE_MAT_OP_SIZE  10
-
-/* maximal linear size of matrix to allocate it on stack. */
-#define  CV_MAX_LOCAL_MAT_SIZE  32
-
-/* maximal size of local memory storage */
-#define  CV_MAX_LOCAL_SIZE  \
-    (CV_MAX_LOCAL_MAT_SIZE*CV_MAX_LOCAL_MAT_SIZE*(int)sizeof(double))
-
-/* default image row align (in bytes) */
-#define  CV_DEFAULT_IMAGE_ROW_ALIGN  4
-
-/* matrices are continuous by default */
-#define  CV_DEFAULT_MAT_ROW_ALIGN  1
-
-/* maximum size of dynamic memory buffer.
-   cvAlloc reports an error if a larger block is requested. */
-#define  CV_MAX_ALLOC_SIZE    (((size_t)1 << (sizeof(size_t)*8-2)))
-
-/* the alignment of all the allocated buffers */
-#define  CV_MALLOC_ALIGN    16
-
-/* default alignment for dynamic data strucutures, resided in storages. */
-#define  CV_STRUCT_ALIGN    ((int)sizeof(double))
-
-/* default storage block size */
-#define  CV_STORAGE_BLOCK_SIZE   ((1<<16) - 128)
-
-/* default memory block for sparse array elements */
-#define  CV_SPARSE_MAT_BLOCK    (1<<12)
-
-/* initial hash table size */
-#define  CV_SPARSE_HASH_SIZE0    (1<<10)
-
-/* maximal average node_count/hash_size ratio beyond which hash table is resized */
-#define  CV_SPARSE_HASH_RATIO    3
-
-/* max length of strings */
-#define  CV_MAX_STRLEN  1024
-
-#if 0 /*def  CV_CHECK_FOR_NANS*/
-#  define CV_CHECK_NANS( arr ) cvCheckArray((arr))
-#else
-#  define CV_CHECK_NANS( arr )
-#endif
-
-/****************************************************************************************\
-*                                  Common declarations                                   *
-\****************************************************************************************/
-
-#ifdef __GNUC__
-#  define CV_DECL_ALIGNED(x) __attribute__ ((aligned (x)))
-#elif defined _MSC_VER
-#  define CV_DECL_ALIGNED(x) __declspec(align(x))
-#else
-#  define CV_DECL_ALIGNED(x)
-#endif
-
-#ifndef CV_IMPL
-#  define CV_IMPL CV_EXTERN_C
-#endif
-
-#define CV_DBG_BREAK() { volatile int* crashMe = 0; *crashMe = 0; }
-
-/* default step, set in case of continuous data
-   to work around checks for valid step in some ipp functions */
-#define  CV_STUB_STEP     (1 << 30)
-
-#define  CV_SIZEOF_FLOAT ((int)sizeof(float))
-#define  CV_SIZEOF_SHORT ((int)sizeof(short))
-
-#define  CV_ORIGIN_TL  0
-#define  CV_ORIGIN_BL  1
-
-/* IEEE754 constants and macros */
-#define  CV_POS_INF       0x7f800000
-#define  CV_NEG_INF       0x807fffff /* CV_TOGGLE_FLT(0xff800000) */
-#define  CV_1F            0x3f800000
-#define  CV_TOGGLE_FLT(x) ((x)^((int)(x) < 0 ? 0x7fffffff : 0))
-#define  CV_TOGGLE_DBL(x) \
-    ((x)^((int64)(x) < 0 ? CV_BIG_INT(0x7fffffffffffffff) : 0))
-
-#define  CV_NOP(a)      (a)
-#define  CV_ADD(a, b)   ((a) + (b))
-#define  CV_SUB(a, b)   ((a) - (b))
-#define  CV_MUL(a, b)   ((a) * (b))
-#define  CV_AND(a, b)   ((a) & (b))
-#define  CV_OR(a, b)    ((a) | (b))
-#define  CV_XOR(a, b)   ((a) ^ (b))
-#define  CV_ANDN(a, b)  (~(a) & (b))
-#define  CV_ORN(a, b)   (~(a) | (b))
-#define  CV_SQR(a)      ((a) * (a))
-
-#define  CV_LT(a, b)    ((a) < (b))
-#define  CV_LE(a, b)    ((a) <= (b))
-#define  CV_EQ(a, b)    ((a) == (b))
-#define  CV_NE(a, b)    ((a) != (b))
-#define  CV_GT(a, b)    ((a) > (b))
-#define  CV_GE(a, b)    ((a) >= (b))
-
-#define  CV_NONZERO(a)      ((a) != 0)
-#define  CV_NONZERO_FLT(a)  (((a)+(a)) != 0)
-
-/* general-purpose saturation macros */
-#define  CV_CAST_8U(t)  (uchar)(!((t) & ~255) ? (t) : (t) > 0 ? 255 : 0)
-#define  CV_CAST_8S(t)  (schar)(!(((t)+128) & ~255) ? (t) : (t) > 0 ? 127 : -128)
-#define  CV_CAST_16U(t) (ushort)(!((t) & ~65535) ? (t) : (t) > 0 ? 65535 : 0)
-#define  CV_CAST_16S(t) (short)(!(((t)+32768) & ~65535) ? (t) : (t) > 0 ? 32767 : -32768)
-#define  CV_CAST_32S(t) (int)(t)
-#define  CV_CAST_64S(t) (int64)(t)
-#define  CV_CAST_32F(t) (float)(t)
-#define  CV_CAST_64F(t) (double)(t)
-
-#define  CV_PASTE2(a,b) a##b
-#define  CV_PASTE(a,b)  CV_PASTE2(a,b)
-
-#define  CV_EMPTY
-#define  CV_MAKE_STR(a) #a
-
-#define  CV_ZERO_OBJ(x) memset((x), 0, sizeof(*(x)))
-
-#define  CV_DIM(static_array) ((int)(sizeof(static_array)/sizeof((static_array)[0])))
-
-#define  cvUnsupportedFormat "Unsupported format"
-
-CV_INLINE void* cvAlignPtr( const void* ptr, int align CV_DEFAULT(32) )
-{
-    assert( (align & (align-1)) == 0 );
-    return (void*)( ((size_t)ptr + align - 1) & ~(size_t)(align-1) );
-}
-
-CV_INLINE int cvAlign( int size, int align )
-{
-    assert( (align & (align-1)) == 0 && size < INT_MAX );
-    return (size + align - 1) & -align;
-}
-
-CV_INLINE  CvSize  cvGetMatSize( const CvMat* mat )
-{
-    CvSize size;
-    size.width = mat->cols;
-    size.height = mat->rows;
-    return size;
-}
-
-#define  CV_DESCALE(x,n)     (((x) + (1 << ((n)-1))) >> (n))
-#define  CV_FLT_TO_FIX(x,n)  cvRound((x)*(1<<(n)))
-
-/****************************************************************************************\
-
-  Generic implementation of QuickSort algorithm.
-  ----------------------------------------------
-  Using this macro user can declare customized sort function that can be much faster
-  than built-in qsort function because of lower overhead on elements
-  comparison and exchange. The macro takes less_than (or LT) argument - a macro or function
-  that takes 2 arguments returns non-zero if the first argument should be before the second
-  one in the sorted sequence and zero otherwise.
-
-  Example:
-
-    Suppose that the task is to sort points by ascending of y coordinates and if
-    y's are equal x's should ascend.
-
-    The code is:
-    ------------------------------------------------------------------------------
-           #define cmp_pts( pt1, pt2 ) \
-               ((pt1).y < (pt2).y || ((pt1).y < (pt2).y && (pt1).x < (pt2).x))
-
-           [static] CV_IMPLEMENT_QSORT( icvSortPoints, CvPoint, cmp_pts )
-    ------------------------------------------------------------------------------
-
-    After that the function "void icvSortPoints( CvPoint* array, size_t total, int aux );"
-    is available to user.
-
-  aux is an additional parameter, which can be used when comparing elements.
-  The current implementation was derived from *BSD system qsort():
-
-    * Copyright (c) 1992, 1993
-    *  The Regents of the University of California.  All rights reserved.
-    *
-    * Redistribution and use in source and binary forms, with or without
-    * modification, are permitted provided that the following conditions
-    * are met:
-    * 1. Redistributions of source code must retain the above copyright
-    *    notice, this list of conditions and the following disclaimer.
-    * 2. Redistributions in binary form must reproduce the above copyright
-    *    notice, this list of conditions and the following disclaimer in the
-    *    documentation and/or other materials provided with the distribution.
-    * 3. All advertising materials mentioning features or use of this software
-    *    must display the following acknowledgement:
-    *  This product includes software developed by the University of
-    *  California, Berkeley and its contributors.
-    * 4. Neither the name of the University nor the names of its contributors
-    *    may be used to endorse or promote products derived from this software
-    *    without specific prior written permission.
-    *
-    * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
-    * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
-    * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
-    * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
-    * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
-    * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
-    * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
-    * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
-    * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
-    * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
-    * SUCH DAMAGE.
-
-\****************************************************************************************/
-
-#define CV_IMPLEMENT_QSORT_EX( func_name, T, LT, user_data_type )                   \
-void func_name( T *array, size_t total, user_data_type aux )                        \
-{                                                                                   \
-    int isort_thresh = 7;                                                           \
-    T t;                                                                            \
-    int sp = 0;                                                                     \
-                                                                                    \
-    struct                                                                          \
-    {                                                                               \
-        T *lb;                                                                      \
-        T *ub;                                                                      \
-    }                                                                               \
-    stack[48];                                                                      \
-                                                                                    \
-    aux = aux;                                                                      \
-                                                                                    \
-    if( total <= 1 )                                                                \
-        return;                                                                     \
-                                                                                    \
-    stack[0].lb = array;                                                            \
-    stack[0].ub = array + (total - 1);                                              \
-                                                                                    \
-    while( sp >= 0 )                                                                \
-    {                                                                               \
-        T* left = stack[sp].lb;                                                     \
-        T* right = stack[sp--].ub;                                                  \
-                                                                                    \
-        for(;;)                                                                     \
-        {                                                                           \
-            int i, n = (int)(right - left) + 1, m;                                  \
-            T* ptr;                                                                 \
-            T* ptr2;                                                                \
-                                                                                    \
-            if( n <= isort_thresh )                                                 \
-            {                                                                       \
-            insert_sort:                                                            \
-                for( ptr = left + 1; ptr <= right; ptr++ )                          \
-                {                                                                   \
-                    for( ptr2 = ptr; ptr2 > left && LT(ptr2[0],ptr2[-1]); ptr2--)   \
-                        CV_SWAP( ptr2[0], ptr2[-1], t );                            \
-                }                                                                   \
-                break;                                                              \
-            }                                                                       \
-            else                                                                    \
-            {                                                                       \
-                T* left0;                                                           \
-                T* left1;                                                           \
-                T* right0;                                                          \
-                T* right1;                                                          \
-                T* pivot;                                                           \
-                T* a;                                                               \
-                T* b;                                                               \
-                T* c;                                                               \
-                int swap_cnt = 0;                                                   \
-                                                                                    \
-                left0 = left;                                                       \
-                right0 = right;                                                     \
-                pivot = left + (n/2);                                               \
-                                                                                    \
-                if( n > 40 )                                                        \
-                {                                                                   \
-                    int d = n / 8;                                                  \
-                    a = left, b = left + d, c = left + 2*d;                         \
-                    left = LT(*a, *b) ? (LT(*b, *c) ? b : (LT(*a, *c) ? c : a))     \
-                                      : (LT(*c, *b) ? b : (LT(*a, *c) ? a : c));    \
-                                                                                    \
-                    a = pivot - d, b = pivot, c = pivot + d;                        \
-                    pivot = LT(*a, *b) ? (LT(*b, *c) ? b : (LT(*a, *c) ? c : a))    \
-                                      : (LT(*c, *b) ? b : (LT(*a, *c) ? a : c));    \
-                                                                                    \
-                    a = right - 2*d, b = right - d, c = right;                      \
-                    right = LT(*a, *b) ? (LT(*b, *c) ? b : (LT(*a, *c) ? c : a))    \
-                                      : (LT(*c, *b) ? b : (LT(*a, *c) ? a : c));    \
-                }                                                                   \
-                                                                                    \
-                a = left, b = pivot, c = right;                                     \
-                pivot = LT(*a, *b) ? (LT(*b, *c) ? b : (LT(*a, *c) ? c : a))        \
-                                   : (LT(*c, *b) ? b : (LT(*a, *c) ? a : c));       \
-                if( pivot != left0 )                                                \
-                {                                                                   \
-                    CV_SWAP( *pivot, *left0, t );                                   \
-                    pivot = left0;                                                  \
-                }                                                                   \
-                left = left1 = left0 + 1;                                           \
-                right = right1 = right0;                                            \
-                                                                                    \
-                for(;;)                                                             \
-                {                                                                   \
-                    while( left <= right && !LT(*pivot, *left) )                    \
-                    {                                                               \
-                        if( !LT(*left, *pivot) )                                    \
-                        {                                                           \
-                            if( left > left1 )                                      \
-                                CV_SWAP( *left1, *left, t );                        \
-                            swap_cnt = 1;                                           \
-                            left1++;                                                \
-                        }                                                           \
-                        left++;                                                     \
-                    }                                                               \
-                                                                                    \
-                    while( left <= right && !LT(*right, *pivot) )                   \
-                    {                                                               \
-                        if( !LT(*pivot, *right) )                                   \
-                        {                                                           \
-                            if( right < right1 )                                    \
-                                CV_SWAP( *right1, *right, t );                      \
-                            swap_cnt = 1;                                           \
-                            right1--;                                               \
-                        }                                                           \
-                        right--;                                                    \
-                    }                                                               \
-                                                                                    \
-                    if( left > right )                                              \
-                        break;                                                      \
-                    CV_SWAP( *left, *right, t );                                    \
-                    swap_cnt = 1;                                                   \
-                    left++;                                                         \
-                    right--;                                                        \
-                }                                                                   \
-                                                                                    \
-                if( swap_cnt == 0 )                                                 \
-                {                                                                   \
-                    left = left0, right = right0;                                   \
-                    goto insert_sort;                                               \
-                }                                                                   \
-                                                                                    \
-                n = MIN( (int)(left1 - left0), (int)(left - left1) );               \
-                for( i = 0; i < n; i++ )                                            \
-                    CV_SWAP( left0[i], left[i-n], t );                              \
-                                                                                    \
-                n = MIN( (int)(right0 - right1), (int)(right1 - right) );           \
-                for( i = 0; i < n; i++ )                                            \
-                    CV_SWAP( left[i], right0[i-n+1], t );                           \
-                n = (int)(left - left1);                                            \
-                m = (int)(right1 - right);                                          \
-                if( n > 1 )                                                         \
-                {                                                                   \
-                    if( m > 1 )                                                     \
-                    {                                                               \
-                        if( n > m )                                                 \
-                        {                                                           \
-                            stack[++sp].lb = left0;                                 \
-                            stack[sp].ub = left0 + n - 1;                           \
-                            left = right0 - m + 1, right = right0;                  \
-                        }                                                           \
-                        else                                                        \
-                        {                                                           \
-                            stack[++sp].lb = right0 - m + 1;                        \
-                            stack[sp].ub = right0;                                  \
-                            left = left0, right = left0 + n - 1;                    \
-                        }                                                           \
-                    }                                                               \
-                    else                                                            \
-                        left = left0, right = left0 + n - 1;                        \
-                }                                                                   \
-                else if( m > 1 )                                                    \
-                    left = right0 - m + 1, right = right0;                          \
-                else                                                                \
-                    break;                                                          \
-            }                                                                       \
-        }                                                                           \
-    }                                                                               \
-}
-
-#define CV_IMPLEMENT_QSORT( func_name, T, cmp )  \
-    CV_IMPLEMENT_QSORT_EX( func_name, T, cmp, int )
-
-/****************************************************************************************\
-*                     Structures and macros for integration with IPP                     *
-\****************************************************************************************/
-
-/* IPP-compatible return codes */
-typedef enum CvStatus
-{
-    CV_BADMEMBLOCK_ERR          = -113,
-    CV_INPLACE_NOT_SUPPORTED_ERR= -112,
-    CV_UNMATCHED_ROI_ERR        = -111,
-    CV_NOTFOUND_ERR             = -110,
-    CV_BADCONVERGENCE_ERR       = -109,
-
-    CV_BADDEPTH_ERR             = -107,
-    CV_BADROI_ERR               = -106,
-    CV_BADHEADER_ERR            = -105,
-    CV_UNMATCHED_FORMATS_ERR    = -104,
-    CV_UNSUPPORTED_COI_ERR      = -103,
-    CV_UNSUPPORTED_CHANNELS_ERR = -102,
-    CV_UNSUPPORTED_DEPTH_ERR    = -101,
-    CV_UNSUPPORTED_FORMAT_ERR   = -100,
-
-    CV_BADARG_ERR               = -49,  //ipp comp
-    CV_NOTDEFINED_ERR           = -48,  //ipp comp
-
-    CV_BADCHANNELS_ERR          = -47,  //ipp comp
-    CV_BADRANGE_ERR             = -44,  //ipp comp
-    CV_BADSTEP_ERR              = -29,  //ipp comp
-
-    CV_BADFLAG_ERR              =  -12,
-    CV_DIV_BY_ZERO_ERR          =  -11, //ipp comp
-    CV_BADCOEF_ERR              =  -10,
-
-    CV_BADFACTOR_ERR            =  -7,
-    CV_BADPOINT_ERR             =  -6,
-    CV_BADSCALE_ERR             =  -4,
-    CV_OUTOFMEM_ERR             =  -3,
-    CV_NULLPTR_ERR              =  -2,
-    CV_BADSIZE_ERR              =  -1,
-    CV_NO_ERR                   =   0,
-    CV_OK                       =   CV_NO_ERR
-}
-CvStatus;
-
-#define CV_NOTHROW throw()
-
-typedef struct CvFuncTable
-{
-    void*   fn_2d[CV_DEPTH_MAX];
-}
-CvFuncTable;
-
-typedef struct CvBigFuncTable
-{
-    void*   fn_2d[CV_DEPTH_MAX*4];
-} CvBigFuncTable;
-
-#define CV_INIT_FUNC_TAB( tab, FUNCNAME, FLAG )         \
-    (tab).fn_2d[CV_8U] = (void*)FUNCNAME##_8u##FLAG;    \
-    (tab).fn_2d[CV_8S] = 0;                             \
-    (tab).fn_2d[CV_16U] = (void*)FUNCNAME##_16u##FLAG;  \
-    (tab).fn_2d[CV_16S] = (void*)FUNCNAME##_16s##FLAG;  \
-    (tab).fn_2d[CV_32S] = (void*)FUNCNAME##_32s##FLAG;  \
-    (tab).fn_2d[CV_32F] = (void*)FUNCNAME##_32f##FLAG;  \
-    (tab).fn_2d[CV_64F] = (void*)FUNCNAME##_64f##FLAG
-
-#ifdef __cplusplus
-
-// < Deprecated
-
-class CV_EXPORTS CvOpenGlFuncTab
-{
-public:
-    virtual ~CvOpenGlFuncTab();
-
-    virtual void genBuffers(int n, unsigned int* buffers) const = 0;
-    virtual void deleteBuffers(int n, const unsigned int* buffers) const = 0;
-
-    virtual void bufferData(unsigned int target, ptrdiff_t size, const void* data, unsigned int usage) const = 0;
-    virtual void bufferSubData(unsigned int target, ptrdiff_t offset, ptrdiff_t size, const void* data) const = 0;
-
-    virtual void bindBuffer(unsigned int target, unsigned int buffer) const = 0;
-
-    virtual void* mapBuffer(unsigned int target, unsigned int access) const = 0;
-    virtual void unmapBuffer(unsigned int target) const = 0;
-
-    virtual void generateBitmapFont(const std::string& family, int height, int weight, bool italic, bool underline, int start, int count, int base) const = 0;
-
-    virtual bool isGlContextInitialized() const = 0;
-};
-
-CV_EXPORTS void icvSetOpenGlFuncTab(const CvOpenGlFuncTab* tab);
-
-CV_EXPORTS bool icvCheckGlError(const char* file, const int line, const char* func = "");
-
-// >
-
-namespace cv { namespace ogl {
-CV_EXPORTS bool checkError(const char* file, const int line, const char* func = "");
-}}
-
-#define CV_CheckGlError() CV_DbgAssert( (cv::ogl::checkError(__FILE__, __LINE__, CV_Func)) )
-
-#endif //__cplusplus
-
-#endif // __OPENCV_CORE_INTERNAL_HPP__

include/opencv2/core/ippasync.hpp

@@ -0,0 +1,195 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                          License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2015, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
+// Copyright (C) 2015, Itseez Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef OPENCV_CORE_IPPASYNC_HPP
+#define OPENCV_CORE_IPPASYNC_HPP
+
+#ifdef HAVE_IPP_A
+
+#include "opencv2/core.hpp"
+#include <ipp_async_op.h>
+#include <ipp_async_accel.h>
+
+namespace cv
+{
+
+namespace hpp
+{
+
+/** @addtogroup core_ipp
+This section describes conversion between OpenCV and [Intel&reg; IPP Asynchronous
+C/C++](http://software.intel.com/en-us/intel-ipp-preview) library. [Getting Started
+Guide](http://registrationcenter.intel.com/irc_nas/3727/ipp_async_get_started.htm) help you to
+install the library, configure header and library build paths.
+ */
+//! @{
+
+    //! convert OpenCV data type to hppDataType
+    inline int toHppType(const int cvType)
+    {
+        int depth = CV_MAT_DEPTH(cvType);
+        int hppType = depth == CV_8U ? HPP_DATA_TYPE_8U :
+                     depth == CV_16U ? HPP_DATA_TYPE_16U :
+                     depth == CV_16S ? HPP_DATA_TYPE_16S :
+                     depth == CV_32S ? HPP_DATA_TYPE_32S :
+                     depth == CV_32F ? HPP_DATA_TYPE_32F :
+                     depth == CV_64F ? HPP_DATA_TYPE_64F : -1;
+        CV_Assert( hppType >= 0 );
+        return hppType;
+    }
+
+    //! convert hppDataType to OpenCV data type
+    inline int toCvType(const int hppType)
+    {
+        int cvType = hppType == HPP_DATA_TYPE_8U ? CV_8U :
+                    hppType == HPP_DATA_TYPE_16U ? CV_16U :
+                    hppType == HPP_DATA_TYPE_16S ? CV_16S :
+                    hppType == HPP_DATA_TYPE_32S ? CV_32S :
+                    hppType == HPP_DATA_TYPE_32F ? CV_32F :
+                    hppType == HPP_DATA_TYPE_64F ? CV_64F : -1;
+        CV_Assert( cvType >= 0 );
+        return cvType;
+    }
+
+    /** @brief Convert hppiMatrix to Mat.
+
+    This function allocates and initializes new matrix (if needed) that has the same size and type as
+    input matrix. Supports CV_8U, CV_16U, CV_16S, CV_32S, CV_32F, CV_64F.
+    @param src input hppiMatrix.
+    @param dst output matrix.
+    @param accel accelerator instance (see hpp::getHpp for the list of acceleration framework types).
+    @param cn number of channels.
+     */
+    inline void copyHppToMat(hppiMatrix* src, Mat& dst, hppAccel accel, int cn)
+    {
+        hppDataType type;
+        hpp32u width, height;
+        hppStatus sts;
+
+        if (src == NULL)
+            return dst.release();
+
+        sts = hppiInquireMatrix(src, &type, &width, &height);
+
+        CV_Assert( sts == HPP_STATUS_NO_ERROR);
+
+        int matType = CV_MAKETYPE(toCvType(type), cn);
+
+        CV_Assert(width%cn == 0);
+
+        width /= cn;
+
+        dst.create((int)height, (int)width, (int)matType);
+
+        size_t newSize = (size_t)(height*(hpp32u)(dst.step));
+
+        sts = hppiGetMatrixData(accel,src,(hpp32u)(dst.step),dst.data,&newSize);
+
+        CV_Assert( sts == HPP_STATUS_NO_ERROR);
+    }
+
+    /** @brief Create Mat from hppiMatrix.
+
+    This function allocates and initializes the Mat that has the same size and type as input matrix.
+    Supports CV_8U, CV_16U, CV_16S, CV_32S, CV_32F, CV_64F.
+    @param src input hppiMatrix.
+    @param accel accelerator instance (see hpp::getHpp for the list of acceleration framework types).
+    @param cn number of channels.
+    @sa howToUseIPPAconversion, hpp::copyHppToMat, hpp::getHpp.
+     */
+    inline Mat getMat(hppiMatrix* src, hppAccel accel, int cn)
+    {
+        Mat dst;
+        copyHppToMat(src, dst, accel, cn);
+        return dst;
+    }
+
+    /** @brief Create hppiMatrix from Mat.
+
+    This function allocates and initializes the hppiMatrix that has the same size and type as input
+    matrix, returns the hppiMatrix*.
+
+    If you want to use zero-copy for GPU you should to have 4KB aligned matrix data. See details
+    [hppiCreateSharedMatrix](http://software.intel.com/ru-ru/node/501697).
+
+    Supports CV_8U, CV_16U, CV_16S, CV_32S, CV_32F, CV_64F.
+
+    @note The hppiMatrix pointer to the image buffer in system memory refers to the src.data. Control
+    the lifetime of the matrix and don't change its data, if there is no special need.
+    @param src input matrix.
+    @param accel accelerator instance. Supports type:
+    -   **HPP_ACCEL_TYPE_CPU** - accelerated by optimized CPU instructions.
+    -   **HPP_ACCEL_TYPE_GPU** - accelerated by GPU programmable units or fixed-function
+        accelerators.
+    -   **HPP_ACCEL_TYPE_ANY** - any acceleration or no acceleration available.
+    @sa howToUseIPPAconversion, hpp::getMat
+     */
+    inline hppiMatrix* getHpp(const Mat& src, hppAccel accel)
+    {
+        int htype = toHppType(src.type());
+        int cn = src.channels();
+
+        CV_Assert(src.data);
+        hppAccelType accelType = hppQueryAccelType(accel);
+
+        if (accelType!=HPP_ACCEL_TYPE_CPU)
+        {
+            hpp32u pitch, size;
+            hppQueryMatrixAllocParams(accel, src.cols*cn, src.rows, htype, &pitch, &size);
+            if (pitch!=0 && size!=0)
+                if ((int)(src.data)%4096==0 && pitch==(hpp32u)(src.step))
+                {
+                    return hppiCreateSharedMatrix(htype, src.cols*cn, src.rows, src.data, pitch, size);
+                }
+        }
+
+        return hppiCreateMatrix(htype, src.cols*cn, src.rows, src.data, (hpp32s)(src.step));;
+    }
+
+//! @}
+}}
+
+#endif
+
+#endif

include/opencv2/core/neon_utils.hpp

@@ -0,0 +1,128 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                          License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2015, Itseez Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef OPENCV_HAL_NEON_UTILS_HPP
+#define OPENCV_HAL_NEON_UTILS_HPP
+
+#include "opencv2/core/cvdef.h"
+
+//! @addtogroup core_utils_neon
+//! @{
+
+#if CV_NEON
+
+inline int32x2_t cv_vrnd_s32_f32(float32x2_t v)
+{
+    static int32x2_t v_sign = vdup_n_s32(1 << 31),
+        v_05 = vreinterpret_s32_f32(vdup_n_f32(0.5f));
+
+    int32x2_t v_addition = vorr_s32(v_05, vand_s32(v_sign, vreinterpret_s32_f32(v)));
+    return vcvt_s32_f32(vadd_f32(v, vreinterpret_f32_s32(v_addition)));
+}
+
+inline int32x4_t cv_vrndq_s32_f32(float32x4_t v)
+{
+    static int32x4_t v_sign = vdupq_n_s32(1 << 31),
+        v_05 = vreinterpretq_s32_f32(vdupq_n_f32(0.5f));
+
+    int32x4_t v_addition = vorrq_s32(v_05, vandq_s32(v_sign, vreinterpretq_s32_f32(v)));
+    return vcvtq_s32_f32(vaddq_f32(v, vreinterpretq_f32_s32(v_addition)));
+}
+
+inline uint32x2_t cv_vrnd_u32_f32(float32x2_t v)
+{
+    static float32x2_t v_05 = vdup_n_f32(0.5f);
+    return vcvt_u32_f32(vadd_f32(v, v_05));
+}
+
+inline uint32x4_t cv_vrndq_u32_f32(float32x4_t v)
+{
+    static float32x4_t v_05 = vdupq_n_f32(0.5f);
+    return vcvtq_u32_f32(vaddq_f32(v, v_05));
+}
+
+inline float32x4_t cv_vrecpq_f32(float32x4_t val)
+{
+    float32x4_t reciprocal = vrecpeq_f32(val);
+    reciprocal = vmulq_f32(vrecpsq_f32(val, reciprocal), reciprocal);
+    reciprocal = vmulq_f32(vrecpsq_f32(val, reciprocal), reciprocal);
+    return reciprocal;
+}
+
+inline float32x2_t cv_vrecp_f32(float32x2_t val)
+{
+    float32x2_t reciprocal = vrecpe_f32(val);
+    reciprocal = vmul_f32(vrecps_f32(val, reciprocal), reciprocal);
+    reciprocal = vmul_f32(vrecps_f32(val, reciprocal), reciprocal);
+    return reciprocal;
+}
+
+inline float32x4_t cv_vrsqrtq_f32(float32x4_t val)
+{
+    float32x4_t e = vrsqrteq_f32(val);
+    e = vmulq_f32(vrsqrtsq_f32(vmulq_f32(e, e), val), e);
+    e = vmulq_f32(vrsqrtsq_f32(vmulq_f32(e, e), val), e);
+    return e;
+}
+
+inline float32x2_t cv_vrsqrt_f32(float32x2_t val)
+{
+    float32x2_t e = vrsqrte_f32(val);
+    e = vmul_f32(vrsqrts_f32(vmul_f32(e, e), val), e);
+    e = vmul_f32(vrsqrts_f32(vmul_f32(e, e), val), e);
+    return e;
+}
+
+inline float32x4_t cv_vsqrtq_f32(float32x4_t val)
+{
+    return cv_vrecpq_f32(cv_vrsqrtq_f32(val));
+}
+
+inline float32x2_t cv_vsqrt_f32(float32x2_t val)
+{
+    return cv_vrecp_f32(cv_vrsqrt_f32(val));
+}
+
+#endif
+
+//! @}
+
+#endif // OPENCV_HAL_NEON_UTILS_HPP

include/opencv2/core/ocl.hpp

@@ -0,0 +1,757 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                           License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the OpenCV Foundation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef OPENCV_OPENCL_HPP
+#define OPENCV_OPENCL_HPP
+
+#include "opencv2/core.hpp"
+
+namespace cv { namespace ocl {
+
+//! @addtogroup core_opencl
+//! @{
+
+CV_EXPORTS_W bool haveOpenCL();
+CV_EXPORTS_W bool useOpenCL();
+CV_EXPORTS_W bool haveAmdBlas();
+CV_EXPORTS_W bool haveAmdFft();
+CV_EXPORTS_W void setUseOpenCL(bool flag);
+CV_EXPORTS_W void finish();
+
+CV_EXPORTS bool haveSVM();
+
+class CV_EXPORTS Context;
+class CV_EXPORTS Device;
+class CV_EXPORTS Kernel;
+class CV_EXPORTS Program;
+class CV_EXPORTS ProgramSource;
+class CV_EXPORTS Queue;
+class CV_EXPORTS PlatformInfo;
+class CV_EXPORTS Image2D;
+
+class CV_EXPORTS Device
+{
+public:
+    Device();
+    explicit Device(void* d);
+    Device(const Device& d);
+    Device& operator = (const Device& d);
+    ~Device();
+
+    void set(void* d);
+
+    enum
+    {
+        TYPE_DEFAULT     = (1 << 0),
+        TYPE_CPU         = (1 << 1),
+        TYPE_GPU         = (1 << 2),
+        TYPE_ACCELERATOR = (1 << 3),
+        TYPE_DGPU        = TYPE_GPU + (1 << 16),
+        TYPE_IGPU        = TYPE_GPU + (1 << 17),
+        TYPE_ALL         = 0xFFFFFFFF
+    };
+
+    String name() const;
+    String extensions() const;
+    String version() const;
+    String vendorName() const;
+    String OpenCL_C_Version() const;
+    String OpenCLVersion() const;
+    int deviceVersionMajor() const;
+    int deviceVersionMinor() const;
+    String driverVersion() const;
+    void* ptr() const;
+
+    int type() const;
+
+    int addressBits() const;
+    bool available() const;
+    bool compilerAvailable() const;
+    bool linkerAvailable() const;
+
+    enum
+    {
+        FP_DENORM=(1 << 0),
+        FP_INF_NAN=(1 << 1),
+        FP_ROUND_TO_NEAREST=(1 << 2),
+        FP_ROUND_TO_ZERO=(1 << 3),
+        FP_ROUND_TO_INF=(1 << 4),
+        FP_FMA=(1 << 5),
+        FP_SOFT_FLOAT=(1 << 6),
+        FP_CORRECTLY_ROUNDED_DIVIDE_SQRT=(1 << 7)
+    };
+    int doubleFPConfig() const;
+    int singleFPConfig() const;
+    int halfFPConfig() const;
+
+    bool endianLittle() const;
+    bool errorCorrectionSupport() const;
+
+    enum
+    {
+        EXEC_KERNEL=(1 << 0),
+        EXEC_NATIVE_KERNEL=(1 << 1)
+    };
+    int executionCapabilities() const;
+
+    size_t globalMemCacheSize() const;
+
+    enum
+    {
+        NO_CACHE=0,
+        READ_ONLY_CACHE=1,
+        READ_WRITE_CACHE=2
+    };
+    int globalMemCacheType() const;
+    int globalMemCacheLineSize() const;
+    size_t globalMemSize() const;
+
+    size_t localMemSize() const;
+    enum
+    {
+        NO_LOCAL_MEM=0,
+        LOCAL_IS_LOCAL=1,
+        LOCAL_IS_GLOBAL=2
+    };
+    int localMemType() const;
+    bool hostUnifiedMemory() const;
+
+    bool imageSupport() const;
+
+    bool imageFromBufferSupport() const;
+    uint imagePitchAlignment() const;
+    uint imageBaseAddressAlignment() const;
+
+    size_t image2DMaxWidth() const;
+    size_t image2DMaxHeight() const;
+
+    size_t image3DMaxWidth() const;
+    size_t image3DMaxHeight() const;
+    size_t image3DMaxDepth() const;
+
+    size_t imageMaxBufferSize() const;
+    size_t imageMaxArraySize() const;
+
+    enum
+    {
+        UNKNOWN_VENDOR=0,
+        VENDOR_AMD=1,
+        VENDOR_INTEL=2,
+        VENDOR_NVIDIA=3
+    };
+    int vendorID() const;
+    // FIXIT
+    // dev.isAMD() doesn't work for OpenCL CPU devices from AMD OpenCL platform.
+    // This method should use platform name instead of vendor name.
+    // After fix restore code in arithm.cpp: ocl_compare()
+    inline bool isAMD() const { return vendorID() == VENDOR_AMD; }
+    inline bool isIntel() const { return vendorID() == VENDOR_INTEL; }
+    inline bool isNVidia() const { return vendorID() == VENDOR_NVIDIA; }
+
+    int maxClockFrequency() const;
+    int maxComputeUnits() const;
+    int maxConstantArgs() const;
+    size_t maxConstantBufferSize() const;
+
+    size_t maxMemAllocSize() const;
+    size_t maxParameterSize() const;
+
+    int maxReadImageArgs() const;
+    int maxWriteImageArgs() const;
+    int maxSamplers() const;
+
+    size_t maxWorkGroupSize() const;
+    int maxWorkItemDims() const;
+    void maxWorkItemSizes(size_t*) const;
+
+    int memBaseAddrAlign() const;
+
+    int nativeVectorWidthChar() const;
+    int nativeVectorWidthShort() const;
+    int nativeVectorWidthInt() const;
+    int nativeVectorWidthLong() const;
+    int nativeVectorWidthFloat() const;
+    int nativeVectorWidthDouble() const;
+    int nativeVectorWidthHalf() const;
+
+    int preferredVectorWidthChar() const;
+    int preferredVectorWidthShort() const;
+    int preferredVectorWidthInt() const;
+    int preferredVectorWidthLong() const;
+    int preferredVectorWidthFloat() const;
+    int preferredVectorWidthDouble() const;
+    int preferredVectorWidthHalf() const;
+
+    size_t printfBufferSize() const;
+    size_t profilingTimerResolution() const;
+
+    static const Device& getDefault();
+
+protected:
+    struct Impl;
+    Impl* p;
+};
+
+
+class CV_EXPORTS Context
+{
+public:
+    Context();
+    explicit Context(int dtype);
+    ~Context();
+    Context(const Context& c);
+    Context& operator = (const Context& c);
+
+    bool create();
+    bool create(int dtype);
+    size_t ndevices() const;
+    const Device& device(size_t idx) const;
+    Program getProg(const ProgramSource& prog,
+                    const String& buildopt, String& errmsg);
+
+    static Context& getDefault(bool initialize = true);
+    void* ptr() const;
+
+    friend void initializeContextFromHandle(Context& ctx, void* platform, void* context, void* device);
+
+    bool useSVM() const;
+    void setUseSVM(bool enabled);
+
+    struct Impl;
+    Impl* p;
+};
+
+class CV_EXPORTS Platform
+{
+public:
+    Platform();
+    ~Platform();
+    Platform(const Platform& p);
+    Platform& operator = (const Platform& p);
+
+    void* ptr() const;
+    static Platform& getDefault();
+
+    friend void initializeContextFromHandle(Context& ctx, void* platform, void* context, void* device);
+protected:
+    struct Impl;
+    Impl* p;
+};
+
+/*
+//! @brief Attaches OpenCL context to OpenCV
+//
+//! @note Note:
+//    OpenCV will check if available OpenCL platform has platformName name,
+//    then assign context to OpenCV and call clRetainContext function.
+//    The deviceID device will be used as target device and new command queue
+//    will be created.
+//
+// Params:
+//! @param platformName - name of OpenCL platform to attach,
+//!                       this string is used to check if platform is available
+//!                       to OpenCV at runtime
+//! @param platfromID   - ID of platform attached context was created for
+//! @param context      - OpenCL context to be attached to OpenCV
+//! @param deviceID     - ID of device, must be created from attached context
+*/
+CV_EXPORTS void attachContext(const String& platformName, void* platformID, void* context, void* deviceID);
+
+/*
+//! @brief Convert OpenCL buffer to UMat
+//
+//! @note Note:
+//   OpenCL buffer (cl_mem_buffer) should contain 2D image data, compatible with OpenCV.
+//   Memory content is not copied from clBuffer to UMat. Instead, buffer handle assigned
+//   to UMat and clRetainMemObject is called.
+//
+// Params:
+//! @param  cl_mem_buffer - source clBuffer handle
+//! @param  step          - num of bytes in single row
+//! @param  rows          - number of rows
+//! @param  cols          - number of cols
+//! @param  type          - OpenCV type of image
+//! @param  dst           - destination UMat
+*/
+CV_EXPORTS void convertFromBuffer(void* cl_mem_buffer, size_t step, int rows, int cols, int type, UMat& dst);
+
+/*
+//! @brief Convert OpenCL image2d_t to UMat
+//
+//! @note Note:
+//   OpenCL image2d_t (cl_mem_image), should be compatible with OpenCV
+//   UMat formats.
+//   Memory content is copied from image to UMat with
+//   clEnqueueCopyImageToBuffer function.
+//
+// Params:
+//! @param  cl_mem_image - source image2d_t handle
+//! @param  dst          - destination UMat
+*/
+CV_EXPORTS void convertFromImage(void* cl_mem_image, UMat& dst);
+
+// TODO Move to internal header
+void initializeContextFromHandle(Context& ctx, void* platform, void* context, void* device);
+
+class CV_EXPORTS Queue
+{
+public:
+    Queue();
+    explicit Queue(const Context& c, const Device& d=Device());
+    ~Queue();
+    Queue(const Queue& q);
+    Queue& operator = (const Queue& q);
+
+    bool create(const Context& c=Context(), const Device& d=Device());
+    void finish();
+    void* ptr() const;
+    static Queue& getDefault();
+
+protected:
+    struct Impl;
+    Impl* p;
+};
+
+
+class CV_EXPORTS KernelArg
+{
+public:
+    enum { LOCAL=1, READ_ONLY=2, WRITE_ONLY=4, READ_WRITE=6, CONSTANT=8, PTR_ONLY = 16, NO_SIZE=256 };
+    KernelArg(int _flags, UMat* _m, int wscale=1, int iwscale=1, const void* _obj=0, size_t _sz=0);
+    KernelArg();
+
+    static KernelArg Local() { return KernelArg(LOCAL, 0); }
+    static KernelArg PtrWriteOnly(const UMat& m)
+    { return KernelArg(PTR_ONLY+WRITE_ONLY, (UMat*)&m); }
+    static KernelArg PtrReadOnly(const UMat& m)
+    { return KernelArg(PTR_ONLY+READ_ONLY, (UMat*)&m); }
+    static KernelArg PtrReadWrite(const UMat& m)
+    { return KernelArg(PTR_ONLY+READ_WRITE, (UMat*)&m); }
+    static KernelArg ReadWrite(const UMat& m, int wscale=1, int iwscale=1)
+    { return KernelArg(READ_WRITE, (UMat*)&m, wscale, iwscale); }
+    static KernelArg ReadWriteNoSize(const UMat& m, int wscale=1, int iwscale=1)
+    { return KernelArg(READ_WRITE+NO_SIZE, (UMat*)&m, wscale, iwscale); }
+    static KernelArg ReadOnly(const UMat& m, int wscale=1, int iwscale=1)
+    { return KernelArg(READ_ONLY, (UMat*)&m, wscale, iwscale); }
+    static KernelArg WriteOnly(const UMat& m, int wscale=1, int iwscale=1)
+    { return KernelArg(WRITE_ONLY, (UMat*)&m, wscale, iwscale); }
+    static KernelArg ReadOnlyNoSize(const UMat& m, int wscale=1, int iwscale=1)
+    { return KernelArg(READ_ONLY+NO_SIZE, (UMat*)&m, wscale, iwscale); }
+    static KernelArg WriteOnlyNoSize(const UMat& m, int wscale=1, int iwscale=1)
+    { return KernelArg(WRITE_ONLY+NO_SIZE, (UMat*)&m, wscale, iwscale); }
+    static KernelArg Constant(const Mat& m);
+    template<typename _Tp> static KernelArg Constant(const _Tp* arr, size_t n)
+    { return KernelArg(CONSTANT, 0, 1, 1, (void*)arr, n); }
+
+    int flags;
+    UMat* m;
+    const void* obj;
+    size_t sz;
+    int wscale, iwscale;
+};
+
+
+class CV_EXPORTS Kernel
+{
+public:
+    Kernel();
+    Kernel(const char* kname, const Program& prog);
+    Kernel(const char* kname, const ProgramSource& prog,
+           const String& buildopts = String(), String* errmsg=0);
+    ~Kernel();
+    Kernel(const Kernel& k);
+    Kernel& operator = (const Kernel& k);
+
+    bool empty() const;
+    bool create(const char* kname, const Program& prog);
+    bool create(const char* kname, const ProgramSource& prog,
+                const String& buildopts, String* errmsg=0);
+
+    int set(int i, const void* value, size_t sz);
+    int set(int i, const Image2D& image2D);
+    int set(int i, const UMat& m);
+    int set(int i, const KernelArg& arg);
+    template<typename _Tp> int set(int i, const _Tp& value)
+    { return set(i, &value, sizeof(value)); }
+
+    template<typename _Tp0>
+    Kernel& args(const _Tp0& a0)
+    {
+        set(0, a0); return *this;
+    }
+
+    template<typename _Tp0, typename _Tp1>
+    Kernel& args(const _Tp0& a0, const _Tp1& a1)
+    {
+        int i = set(0, a0); set(i, a1); return *this;
+    }
+
+    template<typename _Tp0, typename _Tp1, typename _Tp2>
+    Kernel& args(const _Tp0& a0, const _Tp1& a1, const _Tp2& a2)
+    {
+        int i = set(0, a0); i = set(i, a1); set(i, a2); return *this;
+    }
+
+    template<typename _Tp0, typename _Tp1, typename _Tp2, typename _Tp3>
+    Kernel& args(const _Tp0& a0, const _Tp1& a1, const _Tp2& a2, const _Tp3& a3)
+    {
+        int i = set(0, a0); i = set(i, a1); i = set(i, a2); i = set(i, a3); return *this;
+    }
+
+    template<typename _Tp0, typename _Tp1, typename _Tp2, typename _Tp3, typename _Tp4>
+    Kernel& args(const _Tp0& a0, const _Tp1& a1, const _Tp2& a2,
+                 const _Tp3& a3, const _Tp4& a4)
+    {
+        int i = set(0, a0); i = set(i, a1); i = set(i, a2);
+        i = set(i, a3); set(i, a4); return *this;
+    }
+
+    template<typename _Tp0, typename _Tp1, typename _Tp2,
+             typename _Tp3, typename _Tp4, typename _Tp5>
+    Kernel& args(const _Tp0& a0, const _Tp1& a1, const _Tp2& a2,
+                 const _Tp3& a3, const _Tp4& a4, const _Tp5& a5)
+    {
+        int i = set(0, a0); i = set(i, a1); i = set(i, a2);
+        i = set(i, a3); i = set(i, a4); set(i, a5); return *this;
+    }
+
+    template<typename _Tp0, typename _Tp1, typename _Tp2, typename _Tp3,
+             typename _Tp4, typename _Tp5, typename _Tp6>
+    Kernel& args(const _Tp0& a0, const _Tp1& a1, const _Tp2& a2, const _Tp3& a3,
+                 const _Tp4& a4, const _Tp5& a5, const _Tp6& a6)
+    {
+        int i = set(0, a0); i = set(i, a1); i = set(i, a2); i = set(i, a3);
+        i = set(i, a4); i = set(i, a5); set(i, a6); return *this;
+    }
+
+    template<typename _Tp0, typename _Tp1, typename _Tp2, typename _Tp3,
+             typename _Tp4, typename _Tp5, typename _Tp6, typename _Tp7>
+    Kernel& args(const _Tp0& a0, const _Tp1& a1, const _Tp2& a2, const _Tp3& a3,
+                 const _Tp4& a4, const _Tp5& a5, const _Tp6& a6, const _Tp7& a7)
+    {
+        int i = set(0, a0); i = set(i, a1); i = set(i, a2); i = set(i, a3);
+        i = set(i, a4); i = set(i, a5); i = set(i, a6); set(i, a7); return *this;
+    }
+
+    template<typename _Tp0, typename _Tp1, typename _Tp2, typename _Tp3, typename _Tp4,
+             typename _Tp5, typename _Tp6, typename _Tp7, typename _Tp8>
+    Kernel& args(const _Tp0& a0, const _Tp1& a1, const _Tp2& a2, const _Tp3& a3,
+                 const _Tp4& a4, const _Tp5& a5, const _Tp6& a6, const _Tp7& a7,
+                 const _Tp8& a8)
+    {
+        int i = set(0, a0); i = set(i, a1); i = set(i, a2); i = set(i, a3); i = set(i, a4);
+        i = set(i, a5); i = set(i, a6); i = set(i, a7); set(i, a8); return *this;
+    }
+
+    template<typename _Tp0, typename _Tp1, typename _Tp2, typename _Tp3, typename _Tp4,
+             typename _Tp5, typename _Tp6, typename _Tp7, typename _Tp8, typename _Tp9>
+    Kernel& args(const _Tp0& a0, const _Tp1& a1, const _Tp2& a2, const _Tp3& a3,
+                 const _Tp4& a4, const _Tp5& a5, const _Tp6& a6, const _Tp7& a7,
+                 const _Tp8& a8, const _Tp9& a9)
+    {
+        int i = set(0, a0); i = set(i, a1); i = set(i, a2); i = set(i, a3); i = set(i, a4); i = set(i, a5);
+        i = set(i, a6); i = set(i, a7); i = set(i, a8); set(i, a9); return *this;
+    }
+
+    template<typename _Tp0, typename _Tp1, typename _Tp2, typename _Tp3,
+             typename _Tp4, typename _Tp5, typename _Tp6, typename _Tp7,
+             typename _Tp8, typename _Tp9, typename _Tp10>
+    Kernel& args(const _Tp0& a0, const _Tp1& a1, const _Tp2& a2, const _Tp3& a3,
+                 const _Tp4& a4, const _Tp5& a5, const _Tp6& a6, const _Tp7& a7,
+                 const _Tp8& a8, const _Tp9& a9, const _Tp10& a10)
+    {
+        int i = set(0, a0); i = set(i, a1); i = set(i, a2); i = set(i, a3); i = set(i, a4); i = set(i, a5);
+        i = set(i, a6); i = set(i, a7); i = set(i, a8); i = set(i, a9); set(i, a10); return *this;
+    }
+
+    template<typename _Tp0, typename _Tp1, typename _Tp2, typename _Tp3,
+             typename _Tp4, typename _Tp5, typename _Tp6, typename _Tp7,
+             typename _Tp8, typename _Tp9, typename _Tp10, typename _Tp11>
+    Kernel& args(const _Tp0& a0, const _Tp1& a1, const _Tp2& a2, const _Tp3& a3,
+                 const _Tp4& a4, const _Tp5& a5, const _Tp6& a6, const _Tp7& a7,
+                 const _Tp8& a8, const _Tp9& a9, const _Tp10& a10, const _Tp11& a11)
+    {
+        int i = set(0, a0); i = set(i, a1); i = set(i, a2); i = set(i, a3); i = set(i, a4); i = set(i, a5);
+        i = set(i, a6); i = set(i, a7); i = set(i, a8); i = set(i, a9); i = set(i, a10); set(i, a11); return *this;
+    }
+
+    template<typename _Tp0, typename _Tp1, typename _Tp2, typename _Tp3,
+             typename _Tp4, typename _Tp5, typename _Tp6, typename _Tp7,
+             typename _Tp8, typename _Tp9, typename _Tp10, typename _Tp11, typename _Tp12>
+    Kernel& args(const _Tp0& a0, const _Tp1& a1, const _Tp2& a2, const _Tp3& a3,
+                 const _Tp4& a4, const _Tp5& a5, const _Tp6& a6, const _Tp7& a7,
+                 const _Tp8& a8, const _Tp9& a9, const _Tp10& a10, const _Tp11& a11,
+                 const _Tp12& a12)
+    {
+        int i = set(0, a0); i = set(i, a1); i = set(i, a2); i = set(i, a3); i = set(i, a4); i = set(i, a5);
+        i = set(i, a6); i = set(i, a7); i = set(i, a8); i = set(i, a9); i = set(i, a10); i = set(i, a11);
+        set(i, a12); return *this;
+    }
+
+    template<typename _Tp0, typename _Tp1, typename _Tp2, typename _Tp3,
+             typename _Tp4, typename _Tp5, typename _Tp6, typename _Tp7,
+             typename _Tp8, typename _Tp9, typename _Tp10, typename _Tp11, typename _Tp12,
+             typename _Tp13>
+    Kernel& args(const _Tp0& a0, const _Tp1& a1, const _Tp2& a2, const _Tp3& a3,
+                 const _Tp4& a4, const _Tp5& a5, const _Tp6& a6, const _Tp7& a7,
+                 const _Tp8& a8, const _Tp9& a9, const _Tp10& a10, const _Tp11& a11,
+                 const _Tp12& a12, const _Tp13& a13)
+    {
+        int i = set(0, a0); i = set(i, a1); i = set(i, a2); i = set(i, a3); i = set(i, a4); i = set(i, a5);
+        i = set(i, a6); i = set(i, a7); i = set(i, a8); i = set(i, a9); i = set(i, a10); i = set(i, a11);
+        i = set(i, a12); set(i, a13); return *this;
+    }
+
+    template<typename _Tp0, typename _Tp1, typename _Tp2, typename _Tp3,
+             typename _Tp4, typename _Tp5, typename _Tp6, typename _Tp7,
+             typename _Tp8, typename _Tp9, typename _Tp10, typename _Tp11, typename _Tp12,
+             typename _Tp13, typename _Tp14>
+    Kernel& args(const _Tp0& a0, const _Tp1& a1, const _Tp2& a2, const _Tp3& a3,
+                 const _Tp4& a4, const _Tp5& a5, const _Tp6& a6, const _Tp7& a7,
+                 const _Tp8& a8, const _Tp9& a9, const _Tp10& a10, const _Tp11& a11,
+                 const _Tp12& a12, const _Tp13& a13, const _Tp14& a14)
+    {
+        int i = set(0, a0); i = set(i, a1); i = set(i, a2); i = set(i, a3); i = set(i, a4); i = set(i, a5);
+        i = set(i, a6); i = set(i, a7); i = set(i, a8); i = set(i, a9); i = set(i, a10); i = set(i, a11);
+        i = set(i, a12); i = set(i, a13); set(i, a14); return *this;
+    }
+
+    template<typename _Tp0, typename _Tp1, typename _Tp2, typename _Tp3,
+             typename _Tp4, typename _Tp5, typename _Tp6, typename _Tp7,
+             typename _Tp8, typename _Tp9, typename _Tp10, typename _Tp11, typename _Tp12,
+             typename _Tp13, typename _Tp14, typename _Tp15>
+    Kernel& args(const _Tp0& a0, const _Tp1& a1, const _Tp2& a2, const _Tp3& a3,
+                 const _Tp4& a4, const _Tp5& a5, const _Tp6& a6, const _Tp7& a7,
+                 const _Tp8& a8, const _Tp9& a9, const _Tp10& a10, const _Tp11& a11,
+                 const _Tp12& a12, const _Tp13& a13, const _Tp14& a14, const _Tp15& a15)
+    {
+        int i = set(0, a0); i = set(i, a1); i = set(i, a2); i = set(i, a3); i = set(i, a4); i = set(i, a5);
+        i = set(i, a6); i = set(i, a7); i = set(i, a8); i = set(i, a9); i = set(i, a10); i = set(i, a11);
+        i = set(i, a12); i = set(i, a13); i = set(i, a14); set(i, a15); return *this;
+    }
+    /*
+    Run the OpenCL kernel.
+    @param dims the work problem dimensions. It is the length of globalsize and localsize. It can be either 1, 2 or 3.
+    @param globalsize work items for each dimension.
+    It is not the final globalsize passed to OpenCL.
+    Each dimension will be adjusted to the nearest integer divisible by the corresponding value in localsize.
+    If localsize is NULL, it will still be adjusted depending on dims.
+    The adjusted values are greater than or equal to the original values.
+    @param localsize work-group size for each dimension.
+    @param sync specify whether to wait for OpenCL computation to finish before return.
+    @param q command queue
+    */
+    bool run(int dims, size_t globalsize[],
+             size_t localsize[], bool sync, const Queue& q=Queue());
+    bool runTask(bool sync, const Queue& q=Queue());
+
+    size_t workGroupSize() const;
+    size_t preferedWorkGroupSizeMultiple() const;
+    bool compileWorkGroupSize(size_t wsz[]) const;
+    size_t localMemSize() const;
+
+    void* ptr() const;
+    struct Impl;
+
+protected:
+    Impl* p;
+};
+
+class CV_EXPORTS Program
+{
+public:
+    Program();
+    Program(const ProgramSource& src,
+            const String& buildflags, String& errmsg);
+    explicit Program(const String& buf);
+    Program(const Program& prog);
+
+    Program& operator = (const Program& prog);
+    ~Program();
+
+    bool create(const ProgramSource& src,
+                const String& buildflags, String& errmsg);
+    bool read(const String& buf, const String& buildflags);
+    bool write(String& buf) const;
+
+    const ProgramSource& source() const;
+    void* ptr() const;
+
+    String getPrefix() const;
+    static String getPrefix(const String& buildflags);
+
+protected:
+    struct Impl;
+    Impl* p;
+};
+
+
+class CV_EXPORTS ProgramSource
+{
+public:
+    typedef uint64 hash_t;
+
+    ProgramSource();
+    explicit ProgramSource(const String& prog);
+    explicit ProgramSource(const char* prog);
+    ~ProgramSource();
+    ProgramSource(const ProgramSource& prog);
+    ProgramSource& operator = (const ProgramSource& prog);
+
+    const String& source() const;
+    hash_t hash() const;
+
+protected:
+    struct Impl;
+    Impl* p;
+};
+
+class CV_EXPORTS PlatformInfo
+{
+public:
+    PlatformInfo();
+    explicit PlatformInfo(void* id);
+    ~PlatformInfo();
+
+    PlatformInfo(const PlatformInfo& i);
+    PlatformInfo& operator =(const PlatformInfo& i);
+
+    String name() const;
+    String vendor() const;
+    String version() const;
+    int deviceNumber() const;
+    void getDevice(Device& device, int d) const;
+
+protected:
+    struct Impl;
+    Impl* p;
+};
+
+CV_EXPORTS const char* convertTypeStr(int sdepth, int ddepth, int cn, char* buf);
+CV_EXPORTS const char* typeToStr(int t);
+CV_EXPORTS const char* memopTypeToStr(int t);
+CV_EXPORTS const char* vecopTypeToStr(int t);
+CV_EXPORTS String kernelToStr(InputArray _kernel, int ddepth = -1, const char * name = NULL);
+CV_EXPORTS void getPlatfomsInfo(std::vector<PlatformInfo>& platform_info);
+
+
+enum OclVectorStrategy
+{
+    // all matrices have its own vector width
+    OCL_VECTOR_OWN = 0,
+    // all matrices have maximal vector width among all matrices
+    // (useful for cases when matrices have different data types)
+    OCL_VECTOR_MAX = 1,
+
+    // default strategy
+    OCL_VECTOR_DEFAULT = OCL_VECTOR_OWN
+};
+
+CV_EXPORTS int predictOptimalVectorWidth(InputArray src1, InputArray src2 = noArray(), InputArray src3 = noArray(),
+                                         InputArray src4 = noArray(), InputArray src5 = noArray(), InputArray src6 = noArray(),
+                                         InputArray src7 = noArray(), InputArray src8 = noArray(), InputArray src9 = noArray(),
+                                         OclVectorStrategy strat = OCL_VECTOR_DEFAULT);
+
+CV_EXPORTS int checkOptimalVectorWidth(const int *vectorWidths,
+                                       InputArray src1, InputArray src2 = noArray(), InputArray src3 = noArray(),
+                                       InputArray src4 = noArray(), InputArray src5 = noArray(), InputArray src6 = noArray(),
+                                       InputArray src7 = noArray(), InputArray src8 = noArray(), InputArray src9 = noArray(),
+                                       OclVectorStrategy strat = OCL_VECTOR_DEFAULT);
+
+// with OCL_VECTOR_MAX strategy
+CV_EXPORTS int predictOptimalVectorWidthMax(InputArray src1, InputArray src2 = noArray(), InputArray src3 = noArray(),
+                                            InputArray src4 = noArray(), InputArray src5 = noArray(), InputArray src6 = noArray(),
+                                            InputArray src7 = noArray(), InputArray src8 = noArray(), InputArray src9 = noArray());
+
+CV_EXPORTS void buildOptionsAddMatrixDescription(String& buildOptions, const String& name, InputArray _m);
+
+class CV_EXPORTS Image2D
+{
+public:
+    Image2D();
+
+    // src:     The UMat from which to get image properties and data
+    // norm:    Flag to enable the use of normalized channel data types
+    // alias:   Flag indicating that the image should alias the src UMat.
+    //          If true, changes to the image or src will be reflected in
+    //          both objects.
+    explicit Image2D(const UMat &src, bool norm = false, bool alias = false);
+    Image2D(const Image2D & i);
+    ~Image2D();
+
+    Image2D & operator = (const Image2D & i);
+
+    // Indicates if creating an aliased image should succeed.  Depends on the
+    // underlying platform and the dimensions of the UMat.
+    static bool canCreateAlias(const UMat &u);
+
+    // Indicates if the image format is supported.
+    static bool isFormatSupported(int depth, int cn, bool norm);
+
+    void* ptr() const;
+protected:
+    struct Impl;
+    Impl* p;
+};
+
+
+CV_EXPORTS MatAllocator* getOpenCLAllocator();
+
+
+#ifdef __OPENCV_BUILD
+namespace internal {
+
+CV_EXPORTS bool isOpenCLForced();
+#define OCL_FORCE_CHECK(condition) (cv::ocl::internal::isOpenCLForced() || (condition))
+
+CV_EXPORTS bool isPerformanceCheckBypassed();
+#define OCL_PERFORMANCE_CHECK(condition) (cv::ocl::internal::isPerformanceCheckBypassed() || (condition))
+
+CV_EXPORTS bool isCLBuffer(UMat& u);
+
+} // namespace internal
+#endif
+
+//! @}
+
+}}
+
+#endif

include/opencv2/core/ocl_genbase.hpp

@@ -10,8 +10,7 @@
 //                           License Agreement
 //                For Open Source Computer Vision Library
 //
-// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
-// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
 // Third party copyrights are property of their respective owners.
 //
 // Redistribution and use in source and binary forms, with or without modification,
@@ -30,7 +29,7 @@
 // This software is provided by the copyright holders and contributors "as is" and
 // any express or implied warranties, including, but not limited to, the implied
 // warranties of merchantability and fitness for a particular purpose are disclaimed.
-// In no event shall the Intel Corporation or contributors be liable for any direct,
+// In no event shall the OpenCV Foundation or contributors be liable for any direct,
 // indirect, incidental, special, exemplary, or consequential damages
 // (including, but not limited to, procurement of substitute goods or services;
 // loss of use, data, or profits; or business interruption) however caused
@@ -40,4 +39,26 @@
 //
 //M*/
 
-#include "opencv2/core/cuda_devptrs.hpp"
+#ifndef OPENCV_OPENCL_GENBASE_HPP
+#define OPENCV_OPENCL_GENBASE_HPP
+
+namespace cv
+{
+namespace ocl
+{
+
+//! @cond IGNORED
+
+struct ProgramEntry
+{
+    const char* name;
+    const char* programStr;
+    const char* programHash;
+};
+
+//! @endcond
+
+}
+}
+
+#endif

include/opencv2/core/opengl.hpp

@@ -0,0 +1,729 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                           License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef OPENCV_CORE_OPENGL_HPP
+#define OPENCV_CORE_OPENGL_HPP
+
+#ifndef __cplusplus
+#  error opengl.hpp header must be compiled as C++
+#endif
+
+#include "opencv2/core.hpp"
+#include "ocl.hpp"
+
+namespace cv { namespace ogl {
+
+/** @addtogroup core_opengl
+This section describes OpenGL interoperability.
+
+To enable OpenGL support, configure OpenCV using CMake with WITH_OPENGL=ON . Currently OpenGL is
+supported only with WIN32, GTK and Qt backends on Windows and Linux (MacOS and Android are not
+supported). For GTK backend gtkglext-1.0 library is required.
+
+To use OpenGL functionality you should first create OpenGL context (window or frame buffer). You can
+do this with namedWindow function or with other OpenGL toolkit (GLUT, for example).
+*/
+//! @{
+
+/////////////////// OpenGL Objects ///////////////////
+
+/** @brief Smart pointer for OpenGL buffer object with reference counting.
+
+Buffer Objects are OpenGL objects that store an array of unformatted memory allocated by the OpenGL
+context. These can be used to store vertex data, pixel data retrieved from images or the
+framebuffer, and a variety of other things.
+
+ogl::Buffer has interface similar with Mat interface and represents 2D array memory.
+
+ogl::Buffer supports memory transfers between host and device and also can be mapped to CUDA memory.
+ */
+class CV_EXPORTS Buffer
+{
+public:
+    /** @brief The target defines how you intend to use the buffer object.
+    */
+    enum Target
+    {
+        ARRAY_BUFFER         = 0x8892, //!< The buffer will be used as a source for vertex data
+        ELEMENT_ARRAY_BUFFER = 0x8893, //!< The buffer will be used for indices (in glDrawElements, for example)
+        PIXEL_PACK_BUFFER    = 0x88EB, //!< The buffer will be used for reading from OpenGL textures
+        PIXEL_UNPACK_BUFFER  = 0x88EC  //!< The buffer will be used for writing to OpenGL textures
+    };
+
+    enum Access
+    {
+        READ_ONLY  = 0x88B8,
+        WRITE_ONLY = 0x88B9,
+        READ_WRITE = 0x88BA
+    };
+
+    /** @brief The constructors.
+
+    Creates empty ogl::Buffer object, creates ogl::Buffer object from existed buffer ( abufId
+    parameter), allocates memory for ogl::Buffer object or copies from host/device memory.
+     */
+    Buffer();
+
+    /** @overload
+    @param arows Number of rows in a 2D array.
+    @param acols Number of columns in a 2D array.
+    @param atype Array type ( CV_8UC1, ..., CV_64FC4 ). See Mat for details.
+    @param abufId Buffer object name.
+    @param autoRelease Auto release mode (if true, release will be called in object's destructor).
+    */
+    Buffer(int arows, int acols, int atype, unsigned int abufId, bool autoRelease = false);
+
+    /** @overload
+    @param asize 2D array size.
+    @param atype Array type ( CV_8UC1, ..., CV_64FC4 ). See Mat for details.
+    @param abufId Buffer object name.
+    @param autoRelease Auto release mode (if true, release will be called in object's destructor).
+    */
+    Buffer(Size asize, int atype, unsigned int abufId, bool autoRelease = false);
+
+    /** @overload
+    @param arows Number of rows in a 2D array.
+    @param acols Number of columns in a 2D array.
+    @param atype Array type ( CV_8UC1, ..., CV_64FC4 ). See Mat for details.
+    @param target Buffer usage. See cv::ogl::Buffer::Target .
+    @param autoRelease Auto release mode (if true, release will be called in object's destructor).
+    */
+    Buffer(int arows, int acols, int atype, Target target = ARRAY_BUFFER, bool autoRelease = false);
+
+    /** @overload
+    @param asize 2D array size.
+    @param atype Array type ( CV_8UC1, ..., CV_64FC4 ). See Mat for details.
+    @param target Buffer usage. See cv::ogl::Buffer::Target .
+    @param autoRelease Auto release mode (if true, release will be called in object's destructor).
+    */
+    Buffer(Size asize, int atype, Target target = ARRAY_BUFFER, bool autoRelease = false);
+
+    /** @overload
+    @param arr Input array (host or device memory, it can be Mat , cuda::GpuMat or std::vector ).
+    @param target Buffer usage. See cv::ogl::Buffer::Target .
+    @param autoRelease Auto release mode (if true, release will be called in object's destructor).
+    */
+    explicit Buffer(InputArray arr, Target target = ARRAY_BUFFER, bool autoRelease = false);
+
+    /** @brief Allocates memory for ogl::Buffer object.
+
+    @param arows Number of rows in a 2D array.
+    @param acols Number of columns in a 2D array.
+    @param atype Array type ( CV_8UC1, ..., CV_64FC4 ). See Mat for details.
+    @param target Buffer usage. See cv::ogl::Buffer::Target .
+    @param autoRelease Auto release mode (if true, release will be called in object's destructor).
+     */
+    void create(int arows, int acols, int atype, Target target = ARRAY_BUFFER, bool autoRelease = false);
+
+    /** @overload
+    @param asize 2D array size.
+    @param atype Array type ( CV_8UC1, ..., CV_64FC4 ). See Mat for details.
+    @param target Buffer usage. See cv::ogl::Buffer::Target .
+    @param autoRelease Auto release mode (if true, release will be called in object's destructor).
+    */
+    void create(Size asize, int atype, Target target = ARRAY_BUFFER, bool autoRelease = false);
+
+    /** @brief Decrements the reference counter and destroys the buffer object if needed.
+
+    The function will call setAutoRelease(true) .
+     */
+    void release();
+
+    /** @brief Sets auto release mode.
+
+    The lifetime of the OpenGL object is tied to the lifetime of the context. If OpenGL context was
+    bound to a window it could be released at any time (user can close a window). If object's destructor
+    is called after destruction of the context it will cause an error. Thus ogl::Buffer doesn't destroy
+    OpenGL object in destructor by default (all OpenGL resources will be released with OpenGL context).
+    This function can force ogl::Buffer destructor to destroy OpenGL object.
+    @param flag Auto release mode (if true, release will be called in object's destructor).
+     */
+    void setAutoRelease(bool flag);
+
+    /** @brief Copies from host/device memory to OpenGL buffer.
+    @param arr Input array (host or device memory, it can be Mat , cuda::GpuMat or std::vector ).
+    @param target Buffer usage. See cv::ogl::Buffer::Target .
+    @param autoRelease Auto release mode (if true, release will be called in object's destructor).
+     */
+    void copyFrom(InputArray arr, Target target = ARRAY_BUFFER, bool autoRelease = false);
+
+    /** @overload */
+    void copyFrom(InputArray arr, cuda::Stream& stream, Target target = ARRAY_BUFFER, bool autoRelease = false);
+
+    /** @brief Copies from OpenGL buffer to host/device memory or another OpenGL buffer object.
+
+    @param arr Destination array (host or device memory, can be Mat , cuda::GpuMat , std::vector or
+    ogl::Buffer ).
+     */
+    void copyTo(OutputArray arr) const;
+
+    /** @overload */
+    void copyTo(OutputArray arr, cuda::Stream& stream) const;
+
+    /** @brief Creates a full copy of the buffer object and the underlying data.
+
+    @param target Buffer usage for destination buffer.
+    @param autoRelease Auto release mode for destination buffer.
+     */
+    Buffer clone(Target target = ARRAY_BUFFER, bool autoRelease = false) const;
+
+    /** @brief Binds OpenGL buffer to the specified buffer binding point.
+
+    @param target Binding point. See cv::ogl::Buffer::Target .
+     */
+    void bind(Target target) const;
+
+    /** @brief Unbind any buffers from the specified binding point.
+
+    @param target Binding point. See cv::ogl::Buffer::Target .
+     */
+    static void unbind(Target target);
+
+    /** @brief Maps OpenGL buffer to host memory.
+
+    mapHost maps to the client's address space the entire data store of the buffer object. The data can
+    then be directly read and/or written relative to the returned pointer, depending on the specified
+    access policy.
+
+    A mapped data store must be unmapped with ogl::Buffer::unmapHost before its buffer object is used.
+
+    This operation can lead to memory transfers between host and device.
+
+    Only one buffer object can be mapped at a time.
+    @param access Access policy, indicating whether it will be possible to read from, write to, or both
+    read from and write to the buffer object's mapped data store. The symbolic constant must be
+    ogl::Buffer::READ_ONLY , ogl::Buffer::WRITE_ONLY or ogl::Buffer::READ_WRITE .
+     */
+    Mat mapHost(Access access);
+
+    /** @brief Unmaps OpenGL buffer.
+    */
+    void unmapHost();
+
+    //! map to device memory (blocking)
+    cuda::GpuMat mapDevice();
+    void unmapDevice();
+
+    /** @brief Maps OpenGL buffer to CUDA device memory.
+
+    This operatation doesn't copy data. Several buffer objects can be mapped to CUDA memory at a time.
+
+    A mapped data store must be unmapped with ogl::Buffer::unmapDevice before its buffer object is used.
+     */
+    cuda::GpuMat mapDevice(cuda::Stream& stream);
+
+    /** @brief Unmaps OpenGL buffer.
+    */
+    void unmapDevice(cuda::Stream& stream);
+
+    int rows() const;
+    int cols() const;
+    Size size() const;
+    bool empty() const;
+
+    int type() const;
+    int depth() const;
+    int channels() const;
+    int elemSize() const;
+    int elemSize1() const;
+
+    //! get OpenGL opject id
+    unsigned int bufId() const;
+
+    class Impl;
+
+private:
+    Ptr<Impl> impl_;
+    int rows_;
+    int cols_;
+    int type_;
+};
+
+/** @brief Smart pointer for OpenGL 2D texture memory with reference counting.
+ */
+class CV_EXPORTS Texture2D
+{
+public:
+    /** @brief An Image Format describes the way that the images in Textures store their data.
+    */
+    enum Format
+    {
+        NONE            = 0,
+        DEPTH_COMPONENT = 0x1902, //!< Depth
+        RGB             = 0x1907, //!< Red, Green, Blue
+        RGBA            = 0x1908  //!< Red, Green, Blue, Alpha
+    };
+
+    /** @brief The constructors.
+
+    Creates empty ogl::Texture2D object, allocates memory for ogl::Texture2D object or copies from
+    host/device memory.
+     */
+    Texture2D();
+
+    /** @overload */
+    Texture2D(int arows, int acols, Format aformat, unsigned int atexId, bool autoRelease = false);
+
+    /** @overload */
+    Texture2D(Size asize, Format aformat, unsigned int atexId, bool autoRelease = false);
+
+    /** @overload
+    @param arows Number of rows.
+    @param acols Number of columns.
+    @param aformat Image format. See cv::ogl::Texture2D::Format .
+    @param autoRelease Auto release mode (if true, release will be called in object's destructor).
+    */
+    Texture2D(int arows, int acols, Format aformat, bool autoRelease = false);
+
+    /** @overload
+    @param asize 2D array size.
+    @param aformat Image format. See cv::ogl::Texture2D::Format .
+    @param autoRelease Auto release mode (if true, release will be called in object's destructor).
+    */
+    Texture2D(Size asize, Format aformat, bool autoRelease = false);
+
+    /** @overload
+    @param arr Input array (host or device memory, it can be Mat , cuda::GpuMat or ogl::Buffer ).
+    @param autoRelease Auto release mode (if true, release will be called in object's destructor).
+    */
+    explicit Texture2D(InputArray arr, bool autoRelease = false);
+
+    /** @brief Allocates memory for ogl::Texture2D object.
+
+    @param arows Number of rows.
+    @param acols Number of columns.
+    @param aformat Image format. See cv::ogl::Texture2D::Format .
+    @param autoRelease Auto release mode (if true, release will be called in object's destructor).
+     */
+    void create(int arows, int acols, Format aformat, bool autoRelease = false);
+    /** @overload
+    @param asize 2D array size.
+    @param aformat Image format. See cv::ogl::Texture2D::Format .
+    @param autoRelease Auto release mode (if true, release will be called in object's destructor).
+    */
+    void create(Size asize, Format aformat, bool autoRelease = false);
+
+    /** @brief Decrements the reference counter and destroys the texture object if needed.
+
+    The function will call setAutoRelease(true) .
+     */
+    void release();
+
+    /** @brief Sets auto release mode.
+
+    @param flag Auto release mode (if true, release will be called in object's destructor).
+
+    The lifetime of the OpenGL object is tied to the lifetime of the context. If OpenGL context was
+    bound to a window it could be released at any time (user can close a window). If object's destructor
+    is called after destruction of the context it will cause an error. Thus ogl::Texture2D doesn't
+    destroy OpenGL object in destructor by default (all OpenGL resources will be released with OpenGL
+    context). This function can force ogl::Texture2D destructor to destroy OpenGL object.
+     */
+    void setAutoRelease(bool flag);
+
+    /** @brief Copies from host/device memory to OpenGL texture.
+
+    @param arr Input array (host or device memory, it can be Mat , cuda::GpuMat or ogl::Buffer ).
+    @param autoRelease Auto release mode (if true, release will be called in object's destructor).
+     */
+    void copyFrom(InputArray arr, bool autoRelease = false);
+
+    /** @brief Copies from OpenGL texture to host/device memory or another OpenGL texture object.
+
+    @param arr Destination array (host or device memory, can be Mat , cuda::GpuMat , ogl::Buffer or
+    ogl::Texture2D ).
+    @param ddepth Destination depth.
+    @param autoRelease Auto release mode for destination buffer (if arr is OpenGL buffer or texture).
+     */
+    void copyTo(OutputArray arr, int ddepth = CV_32F, bool autoRelease = false) const;
+
+    /** @brief Binds texture to current active texture unit for GL_TEXTURE_2D target.
+    */
+    void bind() const;
+
+    int rows() const;
+    int cols() const;
+    Size size() const;
+    bool empty() const;
+
+    Format format() const;
+
+    //! get OpenGL opject id
+    unsigned int texId() const;
+
+    class Impl;
+
+private:
+    Ptr<Impl> impl_;
+    int rows_;
+    int cols_;
+    Format format_;
+};
+
+/** @brief Wrapper for OpenGL Client-Side Vertex arrays.
+
+ogl::Arrays stores vertex data in ogl::Buffer objects.
+ */
+class CV_EXPORTS Arrays
+{
+public:
+    /** @brief Default constructor
+     */
+    Arrays();
+
+    /** @brief Sets an array of vertex coordinates.
+    @param vertex array with vertex coordinates, can be both host and device memory.
+    */
+    void setVertexArray(InputArray vertex);
+
+    /** @brief Resets vertex coordinates.
+    */
+    void resetVertexArray();
+
+    /** @brief Sets an array of vertex colors.
+    @param color array with vertex colors, can be both host and device memory.
+     */
+    void setColorArray(InputArray color);
+
+    /** @brief Resets vertex colors.
+    */
+    void resetColorArray();
+
+    /** @brief Sets an array of vertex normals.
+    @param normal array with vertex normals, can be both host and device memory.
+     */
+    void setNormalArray(InputArray normal);
+
+    /** @brief Resets vertex normals.
+    */
+    void resetNormalArray();
+
+    /** @brief Sets an array of vertex texture coordinates.
+    @param texCoord array with vertex texture coordinates, can be both host and device memory.
+     */
+    void setTexCoordArray(InputArray texCoord);
+
+    /** @brief Resets vertex texture coordinates.
+    */
+    void resetTexCoordArray();
+
+    /** @brief Releases all inner buffers.
+    */
+    void release();
+
+    /** @brief Sets auto release mode all inner buffers.
+    @param flag Auto release mode.
+     */
+    void setAutoRelease(bool flag);
+
+    /** @brief Binds all vertex arrays.
+    */
+    void bind() const;
+
+    /** @brief Returns the vertex count.
+    */
+    int size() const;
+    bool empty() const;
+
+private:
+    int size_;
+    Buffer vertex_;
+    Buffer color_;
+    Buffer normal_;
+    Buffer texCoord_;
+};
+
+/////////////////// Render Functions ///////////////////
+
+//! render mode
+enum RenderModes {
+    POINTS         = 0x0000,
+    LINES          = 0x0001,
+    LINE_LOOP      = 0x0002,
+    LINE_STRIP     = 0x0003,
+    TRIANGLES      = 0x0004,
+    TRIANGLE_STRIP = 0x0005,
+    TRIANGLE_FAN   = 0x0006,
+    QUADS          = 0x0007,
+    QUAD_STRIP     = 0x0008,
+    POLYGON        = 0x0009
+};
+
+/** @brief Render OpenGL texture or primitives.
+@param tex Texture to draw.
+@param wndRect Region of window, where to draw a texture (normalized coordinates).
+@param texRect Region of texture to draw (normalized coordinates).
+ */
+CV_EXPORTS void render(const Texture2D& tex,
+    Rect_<double> wndRect = Rect_<double>(0.0, 0.0, 1.0, 1.0),
+    Rect_<double> texRect = Rect_<double>(0.0, 0.0, 1.0, 1.0));
+
+/** @overload
+@param arr Array of privitives vertices.
+@param mode Render mode. One of cv::ogl::RenderModes
+@param color Color for all vertices. Will be used if arr doesn't contain color array.
+*/
+CV_EXPORTS void render(const Arrays& arr, int mode = POINTS, Scalar color = Scalar::all(255));
+
+/** @overload
+@param arr Array of privitives vertices.
+@param indices Array of vertices indices (host or device memory).
+@param mode Render mode. One of cv::ogl::RenderModes
+@param color Color for all vertices. Will be used if arr doesn't contain color array.
+*/
+CV_EXPORTS void render(const Arrays& arr, InputArray indices, int mode = POINTS, Scalar color = Scalar::all(255));
+
+/////////////////// CL-GL Interoperability Functions ///////////////////
+
+namespace ocl {
+using namespace cv::ocl;
+
+// TODO static functions in the Context class
+/** @brief Creates OpenCL context from GL.
+@return Returns reference to OpenCL Context
+ */
+CV_EXPORTS Context& initializeContextFromGL();
+
+} // namespace cv::ogl::ocl
+
+/** @brief Converts InputArray to Texture2D object.
+@param src     - source InputArray.
+@param texture - destination Texture2D object.
+ */
+CV_EXPORTS void convertToGLTexture2D(InputArray src, Texture2D& texture);
+
+/** @brief Converts Texture2D object to OutputArray.
+@param texture - source Texture2D object.
+@param dst     - destination OutputArray.
+ */
+CV_EXPORTS void convertFromGLTexture2D(const Texture2D& texture, OutputArray dst);
+
+/** @brief Maps Buffer object to process on CL side (convert to UMat).
+
+Function creates CL buffer from GL one, and then constructs UMat that can be used
+to process buffer data with OpenCV functions. Note that in current implementation
+UMat constructed this way doesn't own corresponding GL buffer object, so it is
+the user responsibility to close down CL/GL buffers relationships by explicitly
+calling unmapGLBuffer() function.
+@param buffer      - source Buffer object.
+@param accessFlags - data access flags (ACCESS_READ|ACCESS_WRITE).
+@return Returns UMat object
+ */
+CV_EXPORTS UMat mapGLBuffer(const Buffer& buffer, int accessFlags = ACCESS_READ|ACCESS_WRITE);
+
+/** @brief Unmaps Buffer object (releases UMat, previously mapped from Buffer).
+
+Function must be called explicitly by the user for each UMat previously constructed
+by the call to mapGLBuffer() function.
+@param u           - source UMat, created by mapGLBuffer().
+ */
+CV_EXPORTS void unmapGLBuffer(UMat& u);
+
+}} // namespace cv::ogl
+
+namespace cv { namespace cuda {
+
+//! @addtogroup cuda
+//! @{
+
+/** @brief Sets a CUDA device and initializes it for the current thread with OpenGL interoperability.
+
+This function should be explicitly called after OpenGL context creation and before any CUDA calls.
+@param device System index of a CUDA device starting with 0.
+@ingroup core_opengl
+ */
+CV_EXPORTS void setGlDevice(int device = 0);
+
+//! @}
+
+}}
+
+//! @cond IGNORED
+
+////////////////////////////////////////////////////////////////////////
+////////////////////////////////////////////////////////////////////////
+////////////////////////////////////////////////////////////////////////
+
+inline
+cv::ogl::Buffer::Buffer(int arows, int acols, int atype, Target target, bool autoRelease) : rows_(0), cols_(0), type_(0)
+{
+    create(arows, acols, atype, target, autoRelease);
+}
+
+inline
+cv::ogl::Buffer::Buffer(Size asize, int atype, Target target, bool autoRelease) : rows_(0), cols_(0), type_(0)
+{
+    create(asize, atype, target, autoRelease);
+}
+
+inline
+void cv::ogl::Buffer::create(Size asize, int atype, Target target, bool autoRelease)
+{
+    create(asize.height, asize.width, atype, target, autoRelease);
+}
+
+inline
+int cv::ogl::Buffer::rows() const
+{
+    return rows_;
+}
+
+inline
+int cv::ogl::Buffer::cols() const
+{
+    return cols_;
+}
+
+inline
+cv::Size cv::ogl::Buffer::size() const
+{
+    return Size(cols_, rows_);
+}
+
+inline
+bool cv::ogl::Buffer::empty() const
+{
+    return rows_ == 0 || cols_ == 0;
+}
+
+inline
+int cv::ogl::Buffer::type() const
+{
+    return type_;
+}
+
+inline
+int cv::ogl::Buffer::depth() const
+{
+    return CV_MAT_DEPTH(type_);
+}
+
+inline
+int cv::ogl::Buffer::channels() const
+{
+    return CV_MAT_CN(type_);
+}
+
+inline
+int cv::ogl::Buffer::elemSize() const
+{
+    return CV_ELEM_SIZE(type_);
+}
+
+inline
+int cv::ogl::Buffer::elemSize1() const
+{
+    return CV_ELEM_SIZE1(type_);
+}
+
+///////
+
+inline
+cv::ogl::Texture2D::Texture2D(int arows, int acols, Format aformat, bool autoRelease) : rows_(0), cols_(0), format_(NONE)
+{
+    create(arows, acols, aformat, autoRelease);
+}
+
+inline
+cv::ogl::Texture2D::Texture2D(Size asize, Format aformat, bool autoRelease) : rows_(0), cols_(0), format_(NONE)
+{
+    create(asize, aformat, autoRelease);
+}
+
+inline
+void cv::ogl::Texture2D::create(Size asize, Format aformat, bool autoRelease)
+{
+    create(asize.height, asize.width, aformat, autoRelease);
+}
+
+inline
+int cv::ogl::Texture2D::rows() const
+{
+    return rows_;
+}
+
+inline
+int cv::ogl::Texture2D::cols() const
+{
+    return cols_;
+}
+
+inline
+cv::Size cv::ogl::Texture2D::size() const
+{
+    return Size(cols_, rows_);
+}
+
+inline
+bool cv::ogl::Texture2D::empty() const
+{
+    return rows_ == 0 || cols_ == 0;
+}
+
+inline
+cv::ogl::Texture2D::Format cv::ogl::Texture2D::format() const
+{
+    return format_;
+}
+
+///////
+
+inline
+cv::ogl::Arrays::Arrays() : size_(0)
+{
+}
+
+inline
+int cv::ogl::Arrays::size() const
+{
+    return size_;
+}
+
+inline
+bool cv::ogl::Arrays::empty() const
+{
+    return size_ == 0;
+}
+
+//! @endcond
+
+#endif /* OPENCV_CORE_OPENGL_HPP */

include/opencv2/core/opengl_interop.hpp

@@ -1,284 +0,0 @@
-/*M///////////////////////////////////////////////////////////////////////////////////////
-//
-//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
-//
-//  By downloading, copying, installing or using the software you agree to this license.
-//  If you do not agree to this license, do not download, install,
-//  copy or use the software.
-//
-//
-//                           License Agreement
-//                For Open Source Computer Vision Library
-//
-// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
-// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
-// Third party copyrights are property of their respective owners.
-//
-// Redistribution and use in source and binary forms, with or without modification,
-// are permitted provided that the following conditions are met:
-//
-//   * Redistribution's of source code must retain the above copyright notice,
-//     this list of conditions and the following disclaimer.
-//
-//   * Redistribution's in binary form must reproduce the above copyright notice,
-//     this list of conditions and the following disclaimer in the documentation
-//     and/or other materials provided with the distribution.
-//
-//   * The name of the copyright holders may not be used to endorse or promote products
-//     derived from this software without specific prior written permission.
-//
-// This software is provided by the copyright holders and contributors "as is" and
-// any express or implied warranties, including, but not limited to, the implied
-// warranties of merchantability and fitness for a particular purpose are disclaimed.
-// In no event shall the Intel Corporation or contributors be liable for any direct,
-// indirect, incidental, special, exemplary, or consequential damages
-// (including, but not limited to, procurement of substitute goods or services;
-// loss of use, data, or profits; or business interruption) however caused
-// and on any theory of liability, whether in contract, strict liability,
-// or tort (including negligence or otherwise) arising in any way out of
-// the use of this software, even if advised of the possibility of such damage.
-//
-//M*/
-
-#ifndef __OPENCV_OPENGL_INTEROP_HPP__
-#define __OPENCV_OPENGL_INTEROP_HPP__
-
-#ifdef __cplusplus
-
-#include "opencv2/core/core.hpp"
-#include "opencv2/core/opengl_interop_deprecated.hpp"
-
-namespace cv { namespace ogl {
-
-/////////////////// OpenGL Objects ///////////////////
-
-//! Smart pointer for OpenGL buffer memory with reference counting.
-class CV_EXPORTS Buffer
-{
-public:
-    enum Target
-    {
-        ARRAY_BUFFER         = 0x8892, //!< The buffer will be used as a source for vertex data
-        ELEMENT_ARRAY_BUFFER = 0x8893, //!< The buffer will be used for indices (in glDrawElements, for example)
-        PIXEL_PACK_BUFFER    = 0x88EB, //!< The buffer will be used for reading from OpenGL textures
-        PIXEL_UNPACK_BUFFER  = 0x88EC  //!< The buffer will be used for writing to OpenGL textures
-    };
-
-    enum Access
-    {
-        READ_ONLY  = 0x88B8,
-        WRITE_ONLY = 0x88B9,
-        READ_WRITE = 0x88BA
-    };
-
-    //! create empty buffer
-    Buffer();
-
-    //! create buffer from existed buffer id
-    Buffer(int arows, int acols, int atype, unsigned int abufId, bool autoRelease = false);
-    Buffer(Size asize, int atype, unsigned int abufId, bool autoRelease = false);
-
-    //! create buffer
-    Buffer(int arows, int acols, int atype, Target target = ARRAY_BUFFER, bool autoRelease = false);
-    Buffer(Size asize, int atype, Target target = ARRAY_BUFFER, bool autoRelease = false);
-
-    //! copy from host/device memory
-    explicit Buffer(InputArray arr, Target target = ARRAY_BUFFER, bool autoRelease = false);
-
-    //! create buffer
-    void create(int arows, int acols, int atype, Target target = ARRAY_BUFFER, bool autoRelease = false);
-    void create(Size asize, int atype, Target target = ARRAY_BUFFER, bool autoRelease = false) { create(asize.height, asize.width, atype, target, autoRelease); }
-
-    //! release memory and delete buffer object
-    void release();
-
-    //! set auto release mode (if true, release will be called in object's destructor)
-    void setAutoRelease(bool flag);
-
-    //! copy from host/device memory
-    void copyFrom(InputArray arr, Target target = ARRAY_BUFFER, bool autoRelease = false);
-
-    //! copy to host/device memory
-    void copyTo(OutputArray arr, Target target = ARRAY_BUFFER, bool autoRelease = false) const;
-
-    //! create copy of current buffer
-    Buffer clone(Target target = ARRAY_BUFFER, bool autoRelease = false) const;
-
-    //! bind buffer for specified target
-    void bind(Target target) const;
-
-    //! unbind any buffers from specified target
-    static void unbind(Target target);
-
-    //! map to host memory
-    Mat mapHost(Access access);
-    void unmapHost();
-
-    //! map to device memory
-    gpu::GpuMat mapDevice();
-    void unmapDevice();
-
-    int rows() const { return rows_; }
-    int cols() const { return cols_; }
-    Size size() const { return Size(cols_, rows_); }
-    bool empty() const { return rows_ == 0 || cols_ == 0; }
-
-    int type() const { return type_; }
-    int depth() const { return CV_MAT_DEPTH(type_); }
-    int channels() const { return CV_MAT_CN(type_); }
-    int elemSize() const { return CV_ELEM_SIZE(type_); }
-    int elemSize1() const { return CV_ELEM_SIZE1(type_); }
-
-    unsigned int bufId() const;
-
-    class Impl;
-
-private:
-    Ptr<Impl> impl_;
-    int rows_;
-    int cols_;
-    int type_;
-};
-
-//! Smart pointer for OpenGL 2D texture memory with reference counting.
-class CV_EXPORTS Texture2D
-{
-public:
-    enum Format
-    {
-        NONE            = 0,
-        DEPTH_COMPONENT = 0x1902, //!< Depth
-        RGB             = 0x1907, //!< Red, Green, Blue
-        RGBA            = 0x1908  //!< Red, Green, Blue, Alpha
-    };
-
-    //! create empty texture
-    Texture2D();
-
-    //! create texture from existed texture id
-    Texture2D(int arows, int acols, Format aformat, unsigned int atexId, bool autoRelease = false);
-    Texture2D(Size asize, Format aformat, unsigned int atexId, bool autoRelease = false);
-
-    //! create texture
-    Texture2D(int arows, int acols, Format aformat, bool autoRelease = false);
-    Texture2D(Size asize, Format aformat, bool autoRelease = false);
-
-    //! copy from host/device memory
-    explicit Texture2D(InputArray arr, bool autoRelease = false);
-
-    //! create texture
-    void create(int arows, int acols, Format aformat, bool autoRelease = false);
-    void create(Size asize, Format aformat, bool autoRelease = false) { create(asize.height, asize.width, aformat, autoRelease); }
-
-    //! release memory and delete texture object
-    void release();
-
-    //! set auto release mode (if true, release will be called in object's destructor)
-    void setAutoRelease(bool flag);
-
-    //! copy from host/device memory
-    void copyFrom(InputArray arr, bool autoRelease = false);
-
-    //! copy to host/device memory
-    void copyTo(OutputArray arr, int ddepth = CV_32F, bool autoRelease = false) const;
-
-    //! bind texture to current active texture unit for GL_TEXTURE_2D target
-    void bind() const;
-
-    int rows() const { return rows_; }
-    int cols() const { return cols_; }
-    Size size() const { return Size(cols_, rows_); }
-    bool empty() const { return rows_ == 0 || cols_ == 0; }
-
-    Format format() const { return format_; }
-
-    unsigned int texId() const;
-
-    class Impl;
-
-private:
-    Ptr<Impl> impl_;
-    int rows_;
-    int cols_;
-    Format format_;
-};
-
-//! OpenGL Arrays
-class CV_EXPORTS Arrays
-{
-public:
-    Arrays();
-
-    void setVertexArray(InputArray vertex);
-    void resetVertexArray();
-
-    void setColorArray(InputArray color);
-    void resetColorArray();
-
-    void setNormalArray(InputArray normal);
-    void resetNormalArray();
-
-    void setTexCoordArray(InputArray texCoord);
-    void resetTexCoordArray();
-
-    void release();
-
-    void setAutoRelease(bool flag);
-
-    void bind() const;
-
-    int size() const { return size_; }
-    bool empty() const { return size_ == 0; }
-
-private:
-    int size_;
-    Buffer vertex_;
-    Buffer color_;
-    Buffer normal_;
-    Buffer texCoord_;
-};
-
-/////////////////// Render Functions ///////////////////
-
-//! render texture rectangle in window
-CV_EXPORTS void render(const Texture2D& tex,
-    Rect_<double> wndRect = Rect_<double>(0.0, 0.0, 1.0, 1.0),
-    Rect_<double> texRect = Rect_<double>(0.0, 0.0, 1.0, 1.0));
-
-//! render mode
-enum {
-    POINTS         = 0x0000,
-    LINES          = 0x0001,
-    LINE_LOOP      = 0x0002,
-    LINE_STRIP     = 0x0003,
-    TRIANGLES      = 0x0004,
-    TRIANGLE_STRIP = 0x0005,
-    TRIANGLE_FAN   = 0x0006,
-    QUADS          = 0x0007,
-    QUAD_STRIP     = 0x0008,
-    POLYGON        = 0x0009
-};
-
-//! render OpenGL arrays
-CV_EXPORTS void render(const Arrays& arr, int mode = POINTS, Scalar color = Scalar::all(255));
-CV_EXPORTS void render(const Arrays& arr, InputArray indices, int mode = POINTS, Scalar color = Scalar::all(255));
-
-}} // namespace cv::gl
-
-namespace cv { namespace gpu {
-
-//! set a CUDA device to use OpenGL interoperability
-CV_EXPORTS void setGlDevice(int device = 0);
-
-}}
-
-namespace cv {
-
-template <> CV_EXPORTS void Ptr<cv::ogl::Buffer::Impl>::delete_obj();
-template <> CV_EXPORTS void Ptr<cv::ogl::Texture2D::Impl>::delete_obj();
-
-}
-
-#endif // __cplusplus
-
-#endif // __OPENCV_OPENGL_INTEROP_HPP__

include/opencv2/core/opengl_interop_deprecated.hpp

@@ -1,330 +0,0 @@
-/*M///////////////////////////////////////////////////////////////////////////////////////
-//
-//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
-//
-//  By downloading, copying, installing or using the software you agree to this license.
-//  If you do not agree to this license, do not download, install,
-//  copy or use the software.
-//
-//
-//                           License Agreement
-//                For Open Source Computer Vision Library
-//
-// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
-// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
-// Third party copyrights are property of their respective owners.
-//
-// Redistribution and use in source and binary forms, with or without modification,
-// are permitted provided that the following conditions are met:
-//
-//   * Redistribution's of source code must retain the above copyright notice,
-//     this list of conditions and the following disclaimer.
-//
-//   * Redistribution's in binary form must reproduce the above copyright notice,
-//     this list of conditions and the following disclaimer in the documentation
-//     and/or other materials provided with the distribution.
-//
-//   * The name of the copyright holders may not be used to endorse or promote products
-//     derived from this software without specific prior written permission.
-//
-// This software is provided by the copyright holders and contributors "as is" and
-// any express or implied warranties, including, but not limited to, the implied
-// warranties of merchantability and fitness for a particular purpose are disclaimed.
-// In no event shall the Intel Corporation or contributors be liable for any direct,
-// indirect, incidental, special, exemplary, or consequential damages
-// (including, but not limited to, procurement of substitute goods or services;
-// loss of use, data, or profits; or business interruption) however caused
-// and on any theory of liability, whether in contract, strict liability,
-// or tort (including negligence or otherwise) arising in any way out of
-// the use of this software, even if advised of the possibility of such damage.
-//
-//M*/
-
-#ifndef __OPENCV_OPENGL_INTEROP_DEPRECATED_HPP__
-#define __OPENCV_OPENGL_INTEROP_DEPRECATED_HPP__
-
-#ifdef __cplusplus
-
-#include "opencv2/core/core.hpp"
-
-namespace cv
-{
-//! Smart pointer for OpenGL buffer memory with reference counting.
-class CV_EXPORTS GlBuffer
-{
-public:
-    enum Usage
-    {
-        ARRAY_BUFFER = 0x8892,  // buffer will use for OpenGL arrays (vertices, colors, normals, etc)
-        TEXTURE_BUFFER = 0x88EC // buffer will ise for OpenGL textures
-    };
-
-    //! create empty buffer
-    explicit GlBuffer(Usage usage);
-
-    //! create buffer
-    GlBuffer(int rows, int cols, int type, Usage usage);
-    GlBuffer(Size size, int type, Usage usage);
-
-    //! copy from host/device memory
-    GlBuffer(InputArray mat, Usage usage);
-
-    void create(int rows, int cols, int type, Usage usage);
-    void create(Size size, int type, Usage usage);
-    void create(int rows, int cols, int type);
-    void create(Size size, int type);
-
-    void release();
-
-    //! copy from host/device memory
-    void copyFrom(InputArray mat);
-
-    void bind() const;
-    void unbind() const;
-
-    //! map to host memory
-    Mat mapHost();
-    void unmapHost();
-
-    //! map to device memory
-    gpu::GpuMat mapDevice();
-    void unmapDevice();
-
-    inline int rows() const { return rows_; }
-    inline int cols() const { return cols_; }
-    inline Size size() const { return Size(cols_, rows_); }
-    inline bool empty() const { return rows_ == 0 || cols_ == 0; }
-
-    inline int type() const { return type_; }
-    inline int depth() const { return CV_MAT_DEPTH(type_); }
-    inline int channels() const { return CV_MAT_CN(type_); }
-    inline int elemSize() const { return CV_ELEM_SIZE(type_); }
-    inline int elemSize1() const { return CV_ELEM_SIZE1(type_); }
-
-    inline Usage usage() const { return usage_; }
-
-    class Impl;
-private:
-    int rows_;
-    int cols_;
-    int type_;
-    Usage usage_;
-
-    Ptr<Impl> impl_;
-};
-
-template <> CV_EXPORTS void Ptr<GlBuffer::Impl>::delete_obj();
-
-//! Smart pointer for OpenGL 2d texture memory with reference counting.
-class CV_EXPORTS GlTexture
-{
-public:
-    //! create empty texture
-    GlTexture();
-
-    //! create texture
-    GlTexture(int rows, int cols, int type);
-    GlTexture(Size size, int type);
-
-    //! copy from host/device memory
-    explicit GlTexture(InputArray mat, bool bgra = true);
-
-    void create(int rows, int cols, int type);
-    void create(Size size, int type);
-    void release();
-
-    //! copy from host/device memory
-    void copyFrom(InputArray mat, bool bgra = true);
-
-    void bind() const;
-    void unbind() const;
-
-    inline int rows() const { return rows_; }
-    inline int cols() const { return cols_; }
-    inline Size size() const { return Size(cols_, rows_); }
-    inline bool empty() const { return rows_ == 0 || cols_ == 0; }
-
-    inline int type() const { return type_; }
-    inline int depth() const { return CV_MAT_DEPTH(type_); }
-    inline int channels() const { return CV_MAT_CN(type_); }
-    inline int elemSize() const { return CV_ELEM_SIZE(type_); }
-    inline int elemSize1() const { return CV_ELEM_SIZE1(type_); }
-
-    class Impl;
-private:
-    int rows_;
-    int cols_;
-    int type_;
-
-    Ptr<Impl> impl_;
-    GlBuffer buf_;
-};
-
-template <> CV_EXPORTS void Ptr<GlTexture::Impl>::delete_obj();
-
-//! OpenGL Arrays
-class CV_EXPORTS GlArrays
-{
-public:
-    inline GlArrays()
-        : vertex_(GlBuffer::ARRAY_BUFFER), color_(GlBuffer::ARRAY_BUFFER), bgra_(true), normal_(GlBuffer::ARRAY_BUFFER), texCoord_(GlBuffer::ARRAY_BUFFER)
-    {
-    }
-
-    void setVertexArray(InputArray vertex);
-    inline void resetVertexArray() { vertex_.release(); }
-
-    void setColorArray(InputArray color, bool bgra = true);
-    inline void resetColorArray() { color_.release(); }
-
-    void setNormalArray(InputArray normal);
-    inline void resetNormalArray() { normal_.release(); }
-
-    void setTexCoordArray(InputArray texCoord);
-    inline void resetTexCoordArray() { texCoord_.release(); }
-
-    void bind() const;
-    void unbind() const;
-
-    inline int rows() const { return vertex_.rows(); }
-    inline int cols() const { return vertex_.cols(); }
-    inline Size size() const { return vertex_.size(); }
-    inline bool empty() const { return vertex_.empty(); }
-
-private:
-    GlBuffer vertex_;
-    GlBuffer color_;
-    bool bgra_;
-    GlBuffer normal_;
-    GlBuffer texCoord_;
-};
-
-//! OpenGL Font
-class CV_EXPORTS GlFont
-{
-public:
-    enum Weight
-    {
-        WEIGHT_LIGHT    = 300,
-        WEIGHT_NORMAL   = 400,
-        WEIGHT_SEMIBOLD = 600,
-        WEIGHT_BOLD     = 700,
-        WEIGHT_BLACK    = 900
-    };
-
-    enum Style
-    {
-        STYLE_NORMAL    = 0,
-        STYLE_ITALIC    = 1,
-        STYLE_UNDERLINE = 2
-    };
-
-    static Ptr<GlFont> get(const std::string& family, int height = 12, Weight weight = WEIGHT_NORMAL, Style style = STYLE_NORMAL);
-
-    void draw(const char* str, int len) const;
-
-    inline const std::string& family() const { return family_; }
-    inline int height() const { return height_; }
-    inline Weight weight() const { return weight_; }
-    inline Style style() const { return style_; }
-
-private:
-    GlFont(const std::string& family, int height, Weight weight, Style style);
-
-    std::string family_;
-    int height_;
-    Weight weight_;
-    Style style_;
-
-    unsigned int base_;
-
-    GlFont(const GlFont&);
-    GlFont& operator =(const GlFont&);
-};
-
-//! render functions
-
-//! render texture rectangle in window
-CV_EXPORTS void render(const GlTexture& tex,
-    Rect_<double> wndRect = Rect_<double>(0.0, 0.0, 1.0, 1.0),
-    Rect_<double> texRect = Rect_<double>(0.0, 0.0, 1.0, 1.0));
-
-//! render mode
-namespace RenderMode {
-    enum {
-        POINTS         = 0x0000,
-        LINES          = 0x0001,
-        LINE_LOOP      = 0x0002,
-        LINE_STRIP     = 0x0003,
-        TRIANGLES      = 0x0004,
-        TRIANGLE_STRIP = 0x0005,
-        TRIANGLE_FAN   = 0x0006,
-        QUADS          = 0x0007,
-        QUAD_STRIP     = 0x0008,
-        POLYGON        = 0x0009
-    };
-}
-
-//! render OpenGL arrays
-CV_EXPORTS void render(const GlArrays& arr, int mode = RenderMode::POINTS, Scalar color = Scalar::all(255));
-
-CV_EXPORTS void render(const std::string& str, const Ptr<GlFont>& font, Scalar color, Point2d pos);
-
-//! OpenGL camera
-class CV_EXPORTS GlCamera
-{
-public:
-    GlCamera();
-
-    void lookAt(Point3d eye, Point3d center, Point3d up);
-    void setCameraPos(Point3d pos, double yaw, double pitch, double roll);
-
-    void setScale(Point3d scale);
-
-    void setProjectionMatrix(const Mat& projectionMatrix, bool transpose = true);
-    void setPerspectiveProjection(double fov, double aspect, double zNear, double zFar);
-    void setOrthoProjection(double left, double right, double bottom, double top, double zNear, double zFar);
-
-    void setupProjectionMatrix() const;
-    void setupModelViewMatrix() const;
-
-private:
-    Point3d eye_;
-    Point3d center_;
-    Point3d up_;
-
-    Point3d pos_;
-    double yaw_;
-    double pitch_;
-    double roll_;
-
-    bool useLookAtParams_;
-
-    Point3d scale_;
-
-    Mat projectionMatrix_;
-
-    double fov_;
-    double aspect_;
-
-    double left_;
-    double right_;
-    double bottom_;
-    double top_;
-
-    double zNear_;
-    double zFar_;
-
-    bool perspectiveProjection_;
-};
-
-inline void GlBuffer::create(Size _size, int _type, Usage _usage) { create(_size.height, _size.width, _type, _usage); }
-inline void GlBuffer::create(int _rows, int _cols, int _type) { create(_rows, _cols, _type, usage()); }
-inline void GlBuffer::create(Size _size, int _type) { create(_size.height, _size.width, _type, usage()); }
-inline void GlTexture::create(Size _size, int _type) { create(_size.height, _size.width, _type); }
-
-} // namespace cv
-
-#endif // __cplusplus
-
-#endif // __OPENCV_OPENGL_INTEROP_DEPRECATED_HPP__

include/opencv2/core/optim.hpp

@@ -0,0 +1,302 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                           License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the OpenCV Foundation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef OPENCV_OPTIM_HPP
+#define OPENCV_OPTIM_HPP
+
+#include "opencv2/core.hpp"
+
+namespace cv
+{
+
+/** @addtogroup core_optim
+The algorithms in this section minimize or maximize function value within specified constraints or
+without any constraints.
+@{
+*/
+
+/** @brief Basic interface for all solvers
+ */
+class CV_EXPORTS MinProblemSolver : public Algorithm
+{
+public:
+    /** @brief Represents function being optimized
+     */
+    class CV_EXPORTS Function
+    {
+    public:
+        virtual ~Function() {}
+        virtual int getDims() const = 0;
+        virtual double getGradientEps() const;
+        virtual double calc(const double* x) const = 0;
+        virtual void getGradient(const double* x,double* grad);
+    };
+
+    /** @brief Getter for the optimized function.
+
+    The optimized function is represented by Function interface, which requires derivatives to
+    implement the sole method calc(double*) to evaluate the function.
+
+    @return Smart-pointer to an object that implements Function interface - it represents the
+    function that is being optimized. It can be empty, if no function was given so far.
+     */
+    virtual Ptr<Function> getFunction() const = 0;
+
+    /** @brief Setter for the optimized function.
+
+    *It should be called at least once before the call to* minimize(), as default value is not usable.
+
+    @param f The new function to optimize.
+     */
+    virtual void setFunction(const Ptr<Function>& f) = 0;
+
+    /** @brief Getter for the previously set terminal criteria for this algorithm.
+
+    @return Deep copy of the terminal criteria used at the moment.
+     */
+    virtual TermCriteria getTermCriteria() const = 0;
+
+    /** @brief Set terminal criteria for solver.
+
+    This method *is not necessary* to be called before the first call to minimize(), as the default
+    value is sensible.
+
+    Algorithm stops when the number of function evaluations done exceeds termcrit.maxCount, when
+    the function values at the vertices of simplex are within termcrit.epsilon range or simplex
+    becomes so small that it can enclosed in a box with termcrit.epsilon sides, whatever comes
+    first.
+    @param termcrit Terminal criteria to be used, represented as cv::TermCriteria structure.
+     */
+    virtual void setTermCriteria(const TermCriteria& termcrit) = 0;
+
+    /** @brief actually runs the algorithm and performs the minimization.
+
+    The sole input parameter determines the centroid of the starting simplex (roughly, it tells
+    where to start), all the others (terminal criteria, initial step, function to be minimized) are
+    supposed to be set via the setters before the call to this method or the default values (not
+    always sensible) will be used.
+
+    @param x The initial point, that will become a centroid of an initial simplex. After the algorithm
+    will terminate, it will be setted to the point where the algorithm stops, the point of possible
+    minimum.
+    @return The value of a function at the point found.
+     */
+    virtual double minimize(InputOutputArray x) = 0;
+};
+
+/** @brief This class is used to perform the non-linear non-constrained minimization of a function,
+
+defined on an `n`-dimensional Euclidean space, using the **Nelder-Mead method**, also known as
+**downhill simplex method**. The basic idea about the method can be obtained from
+<http://en.wikipedia.org/wiki/Nelder-Mead_method>.
+
+It should be noted, that this method, although deterministic, is rather a heuristic and therefore
+may converge to a local minima, not necessary a global one. It is iterative optimization technique,
+which at each step uses an information about the values of a function evaluated only at `n+1`
+points, arranged as a *simplex* in `n`-dimensional space (hence the second name of the method). At
+each step new point is chosen to evaluate function at, obtained value is compared with previous
+ones and based on this information simplex changes it's shape , slowly moving to the local minimum.
+Thus this method is using *only* function values to make decision, on contrary to, say, Nonlinear
+Conjugate Gradient method (which is also implemented in optim).
+
+Algorithm stops when the number of function evaluations done exceeds termcrit.maxCount, when the
+function values at the vertices of simplex are within termcrit.epsilon range or simplex becomes so
+small that it can enclosed in a box with termcrit.epsilon sides, whatever comes first, for some
+defined by user positive integer termcrit.maxCount and positive non-integer termcrit.epsilon.
+
+@note DownhillSolver is a derivative of the abstract interface
+cv::MinProblemSolver, which in turn is derived from the Algorithm interface and is used to
+encapsulate the functionality, common to all non-linear optimization algorithms in the optim
+module.
+
+@note term criteria should meet following condition:
+@code
+    termcrit.type == (TermCriteria::MAX_ITER + TermCriteria::EPS) && termcrit.epsilon > 0 && termcrit.maxCount > 0
+@endcode
+ */
+class CV_EXPORTS DownhillSolver : public MinProblemSolver
+{
+public:
+    /** @brief Returns the initial step that will be used in downhill simplex algorithm.
+
+    @param step Initial step that will be used in algorithm. Note, that although corresponding setter
+    accepts column-vectors as well as row-vectors, this method will return a row-vector.
+    @see DownhillSolver::setInitStep
+     */
+    virtual void getInitStep(OutputArray step) const=0;
+
+    /** @brief Sets the initial step that will be used in downhill simplex algorithm.
+
+    Step, together with initial point (givin in DownhillSolver::minimize) are two `n`-dimensional
+    vectors that are used to determine the shape of initial simplex. Roughly said, initial point
+    determines the position of a simplex (it will become simplex's centroid), while step determines the
+    spread (size in each dimension) of a simplex. To be more precise, if \f$s,x_0\in\mathbb{R}^n\f$ are
+    the initial step and initial point respectively, the vertices of a simplex will be:
+    \f$v_0:=x_0-\frac{1}{2} s\f$ and \f$v_i:=x_0+s_i\f$ for \f$i=1,2,\dots,n\f$ where \f$s_i\f$ denotes
+    projections of the initial step of *n*-th coordinate (the result of projection is treated to be
+    vector given by \f$s_i:=e_i\cdot\left<e_i\cdot s\right>\f$, where \f$e_i\f$ form canonical basis)
+
+    @param step Initial step that will be used in algorithm. Roughly said, it determines the spread
+    (size in each dimension) of an initial simplex.
+     */
+    virtual void setInitStep(InputArray step)=0;
+
+    /** @brief This function returns the reference to the ready-to-use DownhillSolver object.
+
+    All the parameters are optional, so this procedure can be called even without parameters at
+    all. In this case, the default values will be used. As default value for terminal criteria are
+    the only sensible ones, MinProblemSolver::setFunction() and DownhillSolver::setInitStep()
+    should be called upon the obtained object, if the respective parameters were not given to
+    create(). Otherwise, the two ways (give parameters to createDownhillSolver() or miss them out
+    and call the MinProblemSolver::setFunction() and DownhillSolver::setInitStep()) are absolutely
+    equivalent (and will drop the same errors in the same way, should invalid input be detected).
+    @param f Pointer to the function that will be minimized, similarly to the one you submit via
+    MinProblemSolver::setFunction.
+    @param initStep Initial step, that will be used to construct the initial simplex, similarly to the one
+    you submit via MinProblemSolver::setInitStep.
+    @param termcrit Terminal criteria to the algorithm, similarly to the one you submit via
+    MinProblemSolver::setTermCriteria.
+     */
+    static Ptr<DownhillSolver> create(const Ptr<MinProblemSolver::Function>& f=Ptr<MinProblemSolver::Function>(),
+                                      InputArray initStep=Mat_<double>(1,1,0.0),
+                                      TermCriteria termcrit=TermCriteria(TermCriteria::MAX_ITER+TermCriteria::EPS,5000,0.000001));
+};
+
+/** @brief This class is used to perform the non-linear non-constrained minimization of a function
+with known gradient,
+
+defined on an *n*-dimensional Euclidean space, using the **Nonlinear Conjugate Gradient method**.
+The implementation was done based on the beautifully clear explanatory article [An Introduction to
+the Conjugate Gradient Method Without the Agonizing
+Pain](http://www.cs.cmu.edu/~quake-papers/painless-conjugate-gradient.pdf) by Jonathan Richard
+Shewchuk. The method can be seen as an adaptation of a standard Conjugate Gradient method (see, for
+example <http://en.wikipedia.org/wiki/Conjugate_gradient_method>) for numerically solving the
+systems of linear equations.
+
+It should be noted, that this method, although deterministic, is rather a heuristic method and
+therefore may converge to a local minima, not necessary a global one. What is even more disastrous,
+most of its behaviour is ruled by gradient, therefore it essentially cannot distinguish between
+local minima and maxima. Therefore, if it starts sufficiently near to the local maximum, it may
+converge to it. Another obvious restriction is that it should be possible to compute the gradient of
+a function at any point, thus it is preferable to have analytic expression for gradient and
+computational burden should be born by the user.
+
+The latter responsibility is accompilished via the getGradient method of a
+MinProblemSolver::Function interface (which represents function being optimized). This method takes
+point a point in *n*-dimensional space (first argument represents the array of coordinates of that
+point) and comput its gradient (it should be stored in the second argument as an array).
+
+@note class ConjGradSolver thus does not add any new methods to the basic MinProblemSolver interface.
+
+@note term criteria should meet following condition:
+@code
+    termcrit.type == (TermCriteria::MAX_ITER + TermCriteria::EPS) && termcrit.epsilon > 0 && termcrit.maxCount > 0
+    // or
+    termcrit.type == TermCriteria::MAX_ITER) && termcrit.maxCount > 0
+@endcode
+ */
+class CV_EXPORTS ConjGradSolver : public MinProblemSolver
+{
+public:
+    /** @brief This function returns the reference to the ready-to-use ConjGradSolver object.
+
+    All the parameters are optional, so this procedure can be called even without parameters at
+    all. In this case, the default values will be used. As default value for terminal criteria are
+    the only sensible ones, MinProblemSolver::setFunction() should be called upon the obtained
+    object, if the function was not given to create(). Otherwise, the two ways (submit it to
+    create() or miss it out and call the MinProblemSolver::setFunction()) are absolutely equivalent
+    (and will drop the same errors in the same way, should invalid input be detected).
+    @param f Pointer to the function that will be minimized, similarly to the one you submit via
+    MinProblemSolver::setFunction.
+    @param termcrit Terminal criteria to the algorithm, similarly to the one you submit via
+    MinProblemSolver::setTermCriteria.
+    */
+    static Ptr<ConjGradSolver> create(const Ptr<MinProblemSolver::Function>& f=Ptr<ConjGradSolver::Function>(),
+                                      TermCriteria termcrit=TermCriteria(TermCriteria::MAX_ITER+TermCriteria::EPS,5000,0.000001));
+};
+
+//! return codes for cv::solveLP() function
+enum SolveLPResult
+{
+    SOLVELP_UNBOUNDED    = -2, //!< problem is unbounded (target function can achieve arbitrary high values)
+    SOLVELP_UNFEASIBLE    = -1, //!< problem is unfeasible (there are no points that satisfy all the constraints imposed)
+    SOLVELP_SINGLE    = 0, //!< there is only one maximum for target function
+    SOLVELP_MULTI    = 1 //!< there are multiple maxima for target function - the arbitrary one is returned
+};
+
+/** @brief Solve given (non-integer) linear programming problem using the Simplex Algorithm (Simplex Method).
+
+What we mean here by "linear programming problem" (or LP problem, for short) can be formulated as:
+
+\f[\mbox{Maximize } c\cdot x\\
+ \mbox{Subject to:}\\
+ Ax\leq b\\
+ x\geq 0\f]
+
+Where \f$c\f$ is fixed `1`-by-`n` row-vector, \f$A\f$ is fixed `m`-by-`n` matrix, \f$b\f$ is fixed `m`-by-`1`
+column vector and \f$x\f$ is an arbitrary `n`-by-`1` column vector, which satisfies the constraints.
+
+Simplex algorithm is one of many algorithms that are designed to handle this sort of problems
+efficiently. Although it is not optimal in theoretical sense (there exist algorithms that can solve
+any problem written as above in polynomial time, while simplex method degenerates to exponential
+time for some special cases), it is well-studied, easy to implement and is shown to work well for
+real-life purposes.
+
+The particular implementation is taken almost verbatim from **Introduction to Algorithms, third
+edition** by T. H. Cormen, C. E. Leiserson, R. L. Rivest and Clifford Stein. In particular, the
+Bland's rule <http://en.wikipedia.org/wiki/Bland%27s_rule> is used to prevent cycling.
+
+@param Func This row-vector corresponds to \f$c\f$ in the LP problem formulation (see above). It should
+contain 32- or 64-bit floating point numbers. As a convenience, column-vector may be also submitted,
+in the latter case it is understood to correspond to \f$c^T\f$.
+@param Constr `m`-by-`n+1` matrix, whose rightmost column corresponds to \f$b\f$ in formulation above
+and the remaining to \f$A\f$. It should containt 32- or 64-bit floating point numbers.
+@param z The solution will be returned here as a column-vector - it corresponds to \f$c\f$ in the
+formulation above. It will contain 64-bit floating point numbers.
+@return One of cv::SolveLPResult
+ */
+CV_EXPORTS_W int solveLP(const Mat& Func, const Mat& Constr, Mat& z);
+
+//! @}
+
+}// cv
+
+#endif

include/opencv2/core/ovx.hpp

@@ -0,0 +1,28 @@
+// This file is part of OpenCV project.
+// It is subject to the license terms in the LICENSE file found in the top-level directory
+// of this distribution and at http://opencv.org/license.html.
+
+// Copyright (C) 2016, Intel Corporation, all rights reserved.
+// Third party copyrights are property of their respective owners.
+
+// OpenVX related definitions and declarations
+
+#pragma once
+#ifndef OPENCV_OVX_HPP
+#define OPENCV_OVX_HPP
+
+#include "cvdef.h"
+
+namespace cv
+{
+/// Check if use of OpenVX is possible
+CV_EXPORTS_W bool haveOpenVX();
+
+/// Check if use of OpenVX is enabled
+CV_EXPORTS_W bool useOpenVX();
+
+/// Enable/disable use of OpenVX
+CV_EXPORTS_W void setUseOpenVX(bool flag);
+} // namespace cv
+
+#endif // OPENCV_OVX_HPP

include/opencv2/core/private.cuda.hpp

@@ -0,0 +1,172 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                          License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef OPENCV_CORE_PRIVATE_CUDA_HPP
+#define OPENCV_CORE_PRIVATE_CUDA_HPP
+
+#ifndef __OPENCV_BUILD
+#  error this is a private header which should not be used from outside of the OpenCV library
+#endif
+
+#include "cvconfig.h"
+
+#include "opencv2/core/cvdef.h"
+#include "opencv2/core/base.hpp"
+
+#include "opencv2/core/cuda.hpp"
+
+#ifdef HAVE_CUDA
+#  include <cuda.h>
+#  include <cuda_runtime.h>
+#  include <npp.h>
+#  include "opencv2/core/cuda_stream_accessor.hpp"
+#  include "opencv2/core/cuda/common.hpp"
+
+#  define NPP_VERSION (NPP_VERSION_MAJOR * 1000 + NPP_VERSION_MINOR * 100 + NPP_VERSION_BUILD)
+
+#  define CUDART_MINIMUM_REQUIRED_VERSION 6050
+
+#  if (CUDART_VERSION < CUDART_MINIMUM_REQUIRED_VERSION)
+#    error "Insufficient Cuda Runtime library version, please update it."
+#  endif
+
+#  if defined(CUDA_ARCH_BIN_OR_PTX_10)
+#    error "OpenCV CUDA module doesn't support NVIDIA compute capability 1.0"
+#  endif
+#endif
+
+//! @cond IGNORED
+
+namespace cv { namespace cuda {
+    CV_EXPORTS cv::String getNppErrorMessage(int code);
+    CV_EXPORTS cv::String getCudaDriverApiErrorMessage(int code);
+
+    CV_EXPORTS GpuMat getInputMat(InputArray _src, Stream& stream);
+
+    CV_EXPORTS GpuMat getOutputMat(OutputArray _dst, int rows, int cols, int type, Stream& stream);
+    static inline GpuMat getOutputMat(OutputArray _dst, Size size, int type, Stream& stream)
+    {
+        return getOutputMat(_dst, size.height, size.width, type, stream);
+    }
+
+    CV_EXPORTS void syncOutput(const GpuMat& dst, OutputArray _dst, Stream& stream);
+}}
+
+#ifndef HAVE_CUDA
+
+static inline void throw_no_cuda() { CV_Error(cv::Error::GpuNotSupported, "The library is compiled without CUDA support"); }
+
+#else // HAVE_CUDA
+
+static inline void throw_no_cuda() { CV_Error(cv::Error::StsNotImplemented, "The called functionality is disabled for current build or platform"); }
+
+namespace cv { namespace cuda
+{
+    class CV_EXPORTS BufferPool
+    {
+    public:
+        explicit BufferPool(Stream& stream);
+
+        GpuMat getBuffer(int rows, int cols, int type);
+        GpuMat getBuffer(Size size, int type) { return getBuffer(size.height, size.width, type); }
+
+        GpuMat::Allocator* getAllocator() const { return allocator_; }
+
+    private:
+        GpuMat::Allocator* allocator_;
+    };
+
+    static inline void checkNppError(int code, const char* file, const int line, const char* func)
+    {
+        if (code < 0)
+            cv::error(cv::Error::GpuApiCallError, getNppErrorMessage(code), func, file, line);
+    }
+
+    static inline void checkCudaDriverApiError(int code, const char* file, const int line, const char* func)
+    {
+        if (code != CUDA_SUCCESS)
+            cv::error(cv::Error::GpuApiCallError, getCudaDriverApiErrorMessage(code), func, file, line);
+    }
+
+    template<int n> struct NPPTypeTraits;
+    template<> struct NPPTypeTraits<CV_8U>  { typedef Npp8u npp_type; };
+    template<> struct NPPTypeTraits<CV_8S>  { typedef Npp8s npp_type; };
+    template<> struct NPPTypeTraits<CV_16U> { typedef Npp16u npp_type; };
+    template<> struct NPPTypeTraits<CV_16S> { typedef Npp16s npp_type; };
+    template<> struct NPPTypeTraits<CV_32S> { typedef Npp32s npp_type; };
+    template<> struct NPPTypeTraits<CV_32F> { typedef Npp32f npp_type; };
+    template<> struct NPPTypeTraits<CV_64F> { typedef Npp64f npp_type; };
+
+    class NppStreamHandler
+    {
+    public:
+        inline explicit NppStreamHandler(Stream& newStream)
+        {
+            oldStream = nppGetStream();
+            nppSetStream(StreamAccessor::getStream(newStream));
+        }
+
+        inline explicit NppStreamHandler(cudaStream_t newStream)
+        {
+            oldStream = nppGetStream();
+            nppSetStream(newStream);
+        }
+
+        inline ~NppStreamHandler()
+        {
+            nppSetStream(oldStream);
+        }
+
+    private:
+        cudaStream_t oldStream;
+    };
+}}
+
+#define nppSafeCall(expr)  cv::cuda::checkNppError(expr, __FILE__, __LINE__, CV_Func)
+#define cuSafeCall(expr)  cv::cuda::checkCudaDriverApiError(expr, __FILE__, __LINE__, CV_Func)
+
+#endif // HAVE_CUDA
+
+//! @endcond
+
+#endif // OPENCV_CORE_PRIVATE_CUDA_HPP

include/opencv2/core/private.hpp

@@ -0,0 +1,585 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                          License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef OPENCV_CORE_PRIVATE_HPP
+#define OPENCV_CORE_PRIVATE_HPP
+
+#ifndef __OPENCV_BUILD
+#  error this is a private header which should not be used from outside of the OpenCV library
+#endif
+
+#include "opencv2/core.hpp"
+#include "cvconfig.h"
+
+#ifdef HAVE_EIGEN
+#  if defined __GNUC__ && defined __APPLE__
+#    pragma GCC diagnostic ignored "-Wshadow"
+#  endif
+#  include <Eigen/Core>
+#  include "opencv2/core/eigen.hpp"
+#endif
+
+#ifdef HAVE_TBB
+#  include "tbb/tbb.h"
+#  include "tbb/task.h"
+#  undef min
+#  undef max
+#endif
+
+#if defined HAVE_FP16 && (defined __F16C__ || (defined _MSC_VER && _MSC_VER >= 1700))
+#  include <immintrin.h>
+#  define CV_FP16 1
+#elif defined HAVE_FP16 && defined __GNUC__
+#  define CV_FP16 1
+#endif
+
+#ifndef CV_FP16
+#  define CV_FP16 0
+#endif
+
+//! @cond IGNORED
+
+namespace cv
+{
+#ifdef HAVE_TBB
+
+    typedef tbb::blocked_range<int> BlockedRange;
+
+    template<typename Body> static inline
+    void parallel_for( const BlockedRange& range, const Body& body )
+    {
+        tbb::parallel_for(range, body);
+    }
+
+    typedef tbb::split Split;
+
+    template<typename Body> static inline
+    void parallel_reduce( const BlockedRange& range, Body& body )
+    {
+        tbb::parallel_reduce(range, body);
+    }
+
+    typedef tbb::concurrent_vector<Rect> ConcurrentRectVector;
+#else
+    class BlockedRange
+    {
+    public:
+        BlockedRange() : _begin(0), _end(0), _grainsize(0) {}
+        BlockedRange(int b, int e, int g=1) : _begin(b), _end(e), _grainsize(g) {}
+        int begin() const { return _begin; }
+        int end() const { return _end; }
+        int grainsize() const { return _grainsize; }
+
+    protected:
+        int _begin, _end, _grainsize;
+    };
+
+    template<typename Body> static inline
+    void parallel_for( const BlockedRange& range, const Body& body )
+    {
+        body(range);
+    }
+    typedef std::vector<Rect> ConcurrentRectVector;
+
+    class Split {};
+
+    template<typename Body> static inline
+    void parallel_reduce( const BlockedRange& range, Body& body )
+    {
+        body(range);
+    }
+#endif
+
+    // Returns a static string if there is a parallel framework,
+    // NULL otherwise.
+    CV_EXPORTS const char* currentParallelFramework();
+} //namespace cv
+
+/****************************************************************************************\
+*                                  Common declarations                                   *
+\****************************************************************************************/
+
+/* the alignment of all the allocated buffers */
+#define  CV_MALLOC_ALIGN    16
+
+/* IEEE754 constants and macros */
+#define  CV_TOGGLE_FLT(x) ((x)^((int)(x) < 0 ? 0x7fffffff : 0))
+#define  CV_TOGGLE_DBL(x) ((x)^((int64)(x) < 0 ? CV_BIG_INT(0x7fffffffffffffff) : 0))
+
+static inline void* cvAlignPtr( const void* ptr, int align = 32 )
+{
+    CV_DbgAssert ( (align & (align-1)) == 0 );
+    return (void*)( ((size_t)ptr + align - 1) & ~(size_t)(align-1) );
+}
+
+static inline int cvAlign( int size, int align )
+{
+    CV_DbgAssert( (align & (align-1)) == 0 && size < INT_MAX );
+    return (size + align - 1) & -align;
+}
+
+#ifdef IPL_DEPTH_8U
+static inline cv::Size cvGetMatSize( const CvMat* mat )
+{
+    return cv::Size(mat->cols, mat->rows);
+}
+#endif
+
+namespace cv
+{
+CV_EXPORTS void scalarToRawData(const cv::Scalar& s, void* buf, int type, int unroll_to = 0);
+}
+
+// property implementation macros
+
+#define CV_IMPL_PROPERTY_RO(type, name, member) \
+    inline type get##name() const { return member; }
+
+#define CV_HELP_IMPL_PROPERTY(r_type, w_type, name, member) \
+    CV_IMPL_PROPERTY_RO(r_type, name, member) \
+    inline void set##name(w_type val) { member = val; }
+
+#define CV_HELP_WRAP_PROPERTY(r_type, w_type, name, internal_name, internal_obj) \
+    r_type get##name() const { return internal_obj.get##internal_name(); } \
+    void set##name(w_type val) { internal_obj.set##internal_name(val); }
+
+#define CV_IMPL_PROPERTY(type, name, member) CV_HELP_IMPL_PROPERTY(type, type, name, member)
+#define CV_IMPL_PROPERTY_S(type, name, member) CV_HELP_IMPL_PROPERTY(type, const type &, name, member)
+
+#define CV_WRAP_PROPERTY(type, name, internal_name, internal_obj)  CV_HELP_WRAP_PROPERTY(type, type, name, internal_name, internal_obj)
+#define CV_WRAP_PROPERTY_S(type, name, internal_name, internal_obj) CV_HELP_WRAP_PROPERTY(type, const type &, name, internal_name, internal_obj)
+
+#define CV_WRAP_SAME_PROPERTY(type, name, internal_obj) CV_WRAP_PROPERTY(type, name, name, internal_obj)
+#define CV_WRAP_SAME_PROPERTY_S(type, name, internal_obj) CV_WRAP_PROPERTY_S(type, name, name, internal_obj)
+
+/****************************************************************************************\
+*                     Structures and macros for integration with IPP                     *
+\****************************************************************************************/
+
+#ifdef HAVE_IPP
+#include "ipp.h"
+
+#ifndef IPP_VERSION_UPDATE // prior to 7.1
+#define IPP_VERSION_UPDATE 0
+#endif
+
+#define IPP_VERSION_X100 (IPP_VERSION_MAJOR * 100 + IPP_VERSION_MINOR*10 + IPP_VERSION_UPDATE)
+
+// General define for ipp function disabling
+#define IPP_DISABLE_BLOCK 0
+
+#ifdef CV_MALLOC_ALIGN
+#undef CV_MALLOC_ALIGN
+#endif
+#define CV_MALLOC_ALIGN 32 // required for AVX optimization
+
+#define setIppErrorStatus() cv::ipp::setIppStatus(-1, CV_Func, __FILE__, __LINE__)
+
+static inline IppiSize ippiSize(int width, int height)
+{
+    IppiSize size = { width, height };
+    return size;
+}
+
+static inline IppiSize ippiSize(const cv::Size & _size)
+{
+    IppiSize size = { _size.width, _size.height };
+    return size;
+}
+
+static inline IppiPoint ippiPoint(const cv::Point & _point)
+{
+    IppiPoint point = { _point.x, _point.y };
+    return point;
+}
+
+static inline IppiPoint ippiPoint(int x, int y)
+{
+    IppiPoint point = { x, y };
+    return point;
+}
+
+static inline IppiBorderType ippiGetBorderType(int borderTypeNI)
+{
+    return borderTypeNI == cv::BORDER_CONSTANT ? ippBorderConst :
+        borderTypeNI == cv::BORDER_WRAP ? ippBorderWrap :
+        borderTypeNI == cv::BORDER_REPLICATE ? ippBorderRepl :
+        borderTypeNI == cv::BORDER_REFLECT_101 ? ippBorderMirror :
+        borderTypeNI == cv::BORDER_REFLECT ? ippBorderMirrorR : (IppiBorderType)-1;
+}
+
+static inline IppDataType ippiGetDataType(int depth)
+{
+    return depth == CV_8U ? ipp8u :
+        depth == CV_8S ? ipp8s :
+        depth == CV_16U ? ipp16u :
+        depth == CV_16S ? ipp16s :
+        depth == CV_32S ? ipp32s :
+        depth == CV_32F ? ipp32f :
+        depth == CV_64F ? ipp64f : (IppDataType)-1;
+}
+
+// IPP temporary buffer hepler
+template<typename T>
+class IppAutoBuffer
+{
+public:
+    IppAutoBuffer() { m_pBuffer = NULL; }
+    IppAutoBuffer(int size) { Alloc(size); }
+    ~IppAutoBuffer() { Release(); }
+    T* Alloc(int size) { m_pBuffer = (T*)ippMalloc(size); return m_pBuffer; }
+    void Release() { if(m_pBuffer) ippFree(m_pBuffer); }
+    inline operator T* () { return (T*)m_pBuffer;}
+    inline operator const T* () const { return (const T*)m_pBuffer;}
+private:
+    // Disable copy operations
+    IppAutoBuffer(IppAutoBuffer &) {}
+    IppAutoBuffer& operator =(const IppAutoBuffer &) {return *this;}
+
+    T* m_pBuffer;
+};
+
+#else
+#define IPP_VERSION_X100 0
+#endif
+
+#if defined HAVE_IPP
+#if IPP_VERSION_X100 >= 900
+#define IPP_INITIALIZER(FEAT)                           \
+{                                                       \
+    if(FEAT)                                            \
+        ippSetCpuFeatures(FEAT);                        \
+    else                                                \
+        ippInit();                                      \
+}
+#elif IPP_VERSION_X100 >= 800
+#define IPP_INITIALIZER(FEAT)                           \
+{                                                       \
+    ippInit();                                          \
+}
+#else
+#define IPP_INITIALIZER(FEAT)                           \
+{                                                       \
+    ippStaticInit();                                    \
+}
+#endif
+
+#ifdef CVAPI_EXPORTS
+#define IPP_INITIALIZER_AUTO                            \
+struct __IppInitializer__                               \
+{                                                       \
+    __IppInitializer__()                                \
+    {IPP_INITIALIZER(cv::ipp::getIppFeatures())}        \
+};                                                      \
+static struct __IppInitializer__ __ipp_initializer__;
+#else
+#define IPP_INITIALIZER_AUTO
+#endif
+#else
+#define IPP_INITIALIZER
+#define IPP_INITIALIZER_AUTO
+#endif
+
+#define CV_IPP_CHECK_COND (cv::ipp::useIPP())
+#define CV_IPP_CHECK() if(CV_IPP_CHECK_COND)
+
+#ifdef HAVE_IPP
+
+#ifdef CV_IPP_RUN_VERBOSE
+#define CV_IPP_RUN_(condition, func, ...)                                   \
+    {                                                                       \
+        if (cv::ipp::useIPP() && (condition) && (func))                     \
+        {                                                                   \
+            printf("%s: IPP implementation is running\n", CV_Func);         \
+            fflush(stdout);                                                 \
+            CV_IMPL_ADD(CV_IMPL_IPP);                                       \
+            return __VA_ARGS__;                                             \
+        }                                                                   \
+        else                                                                \
+        {                                                                   \
+            printf("%s: Plain implementation is running\n", CV_Func);       \
+            fflush(stdout);                                                 \
+        }                                                                   \
+    }
+#elif defined CV_IPP_RUN_ASSERT
+#define CV_IPP_RUN_(condition, func, ...)                                   \
+    {                                                                       \
+        if (cv::ipp::useIPP() && (condition))                               \
+        {                                                                   \
+            if(func)                                                        \
+            {                                                               \
+                CV_IMPL_ADD(CV_IMPL_IPP);                                   \
+            }                                                               \
+            else                                                            \
+            {                                                               \
+                setIppErrorStatus();                                        \
+                CV_Error(cv::Error::StsAssert, #func);                      \
+            }                                                               \
+            return __VA_ARGS__;                                             \
+        }                                                                   \
+    }
+#else
+#define CV_IPP_RUN_(condition, func, ...)                                   \
+    if (cv::ipp::useIPP() && (condition) && (func))                         \
+    {                                                                       \
+        CV_IMPL_ADD(CV_IMPL_IPP);                                           \
+        return __VA_ARGS__;                                                 \
+    }
+#endif
+#define CV_IPP_RUN_FAST(func, ...)                                          \
+    if (cv::ipp::useIPP() && (func))                                        \
+    {                                                                       \
+        CV_IMPL_ADD(CV_IMPL_IPP);                                           \
+        return __VA_ARGS__;                                                 \
+    }
+#else
+#define CV_IPP_RUN_(condition, func, ...)
+#define CV_IPP_RUN_FAST(func, ...)
+#endif
+
+#define CV_IPP_RUN(condition, func, ...) CV_IPP_RUN_((condition), (func), __VA_ARGS__)
+
+
+#ifndef IPPI_CALL
+#  define IPPI_CALL(func) CV_Assert((func) >= 0)
+#endif
+
+/* IPP-compatible return codes */
+typedef enum CvStatus
+{
+    CV_BADMEMBLOCK_ERR          = -113,
+    CV_INPLACE_NOT_SUPPORTED_ERR= -112,
+    CV_UNMATCHED_ROI_ERR        = -111,
+    CV_NOTFOUND_ERR             = -110,
+    CV_BADCONVERGENCE_ERR       = -109,
+
+    CV_BADDEPTH_ERR             = -107,
+    CV_BADROI_ERR               = -106,
+    CV_BADHEADER_ERR            = -105,
+    CV_UNMATCHED_FORMATS_ERR    = -104,
+    CV_UNSUPPORTED_COI_ERR      = -103,
+    CV_UNSUPPORTED_CHANNELS_ERR = -102,
+    CV_UNSUPPORTED_DEPTH_ERR    = -101,
+    CV_UNSUPPORTED_FORMAT_ERR   = -100,
+
+    CV_BADARG_ERR               = -49,  //ipp comp
+    CV_NOTDEFINED_ERR           = -48,  //ipp comp
+
+    CV_BADCHANNELS_ERR          = -47,  //ipp comp
+    CV_BADRANGE_ERR             = -44,  //ipp comp
+    CV_BADSTEP_ERR              = -29,  //ipp comp
+
+    CV_BADFLAG_ERR              =  -12,
+    CV_DIV_BY_ZERO_ERR          =  -11, //ipp comp
+    CV_BADCOEF_ERR              =  -10,
+
+    CV_BADFACTOR_ERR            =  -7,
+    CV_BADPOINT_ERR             =  -6,
+    CV_BADSCALE_ERR             =  -4,
+    CV_OUTOFMEM_ERR             =  -3,
+    CV_NULLPTR_ERR              =  -2,
+    CV_BADSIZE_ERR              =  -1,
+    CV_NO_ERR                   =   0,
+    CV_OK                       =   CV_NO_ERR
+}
+CvStatus;
+
+#ifdef HAVE_TEGRA_OPTIMIZATION
+namespace tegra {
+
+CV_EXPORTS bool useTegra();
+CV_EXPORTS void setUseTegra(bool flag);
+
+}
+#endif
+
+#ifdef ENABLE_INSTRUMENTATION
+namespace cv
+{
+namespace instr
+{
+struct InstrTLSStruct
+{
+    InstrTLSStruct()
+    {
+        pCurrentNode = NULL;
+    }
+    InstrNode* pCurrentNode;
+};
+
+class InstrStruct
+{
+public:
+    InstrStruct()
+    {
+        useInstr    = false;
+        flags       = FLAGS_MAPPING;
+        maxDepth    = 0;
+
+        rootNode.m_payload = NodeData("ROOT", NULL, 0, NULL, false, TYPE_GENERAL, IMPL_PLAIN);
+        tlsStruct.get()->pCurrentNode = &rootNode;
+    }
+
+    Mutex mutexCreate;
+    Mutex mutexCount;
+
+    bool       useInstr;
+    int        flags;
+    int        maxDepth;
+    InstrNode  rootNode;
+    TLSData<InstrTLSStruct> tlsStruct;
+};
+
+class CV_EXPORTS IntrumentationRegion
+{
+public:
+    IntrumentationRegion(const char* funName, const char* fileName, int lineNum, void *retAddress, bool alwaysExpand, TYPE instrType = TYPE_GENERAL, IMPL implType = IMPL_PLAIN);
+    ~IntrumentationRegion();
+
+private:
+    bool    m_disabled; // region status
+    uint64  m_regionTicks;
+};
+
+CV_EXPORTS InstrStruct& getInstrumentStruct();
+InstrTLSStruct&         getInstrumentTLSStruct();
+CV_EXPORTS InstrNode*   getCurrentNode();
+}
+}
+
+#ifdef _WIN32
+#define CV_INSTRUMENT_GET_RETURN_ADDRESS _ReturnAddress()
+#else
+#define CV_INSTRUMENT_GET_RETURN_ADDRESS __builtin_extract_return_addr(__builtin_return_address(0))
+#endif
+
+// Instrument region
+#define CV_INSTRUMENT_REGION_META(NAME, ALWAYS_EXPAND, TYPE, IMPL)        ::cv::instr::IntrumentationRegion __instr_region__(NAME, __FILE__, __LINE__, CV_INSTRUMENT_GET_RETURN_ADDRESS, ALWAYS_EXPAND, TYPE, IMPL);
+#define CV_INSTRUMENT_REGION_CUSTOM_META(NAME, ALWAYS_EXPAND, TYPE, IMPL)\
+    void *__curr_address__ = [&]() {return CV_INSTRUMENT_GET_RETURN_ADDRESS;}();\
+    ::cv::instr::IntrumentationRegion __instr_region__(NAME, __FILE__, __LINE__, __curr_address__, false, ::cv::instr::TYPE_GENERAL, ::cv::instr::IMPL_PLAIN);
+// Instrument functions with non-void return type
+#define CV_INSTRUMENT_FUN_RT_META(TYPE, IMPL, ERROR_COND, FUN, ...) ([&]()\
+{\
+    if(::cv::instr::useInstrumentation()){\
+        ::cv::instr::IntrumentationRegion __instr__(#FUN, __FILE__, __LINE__, NULL, false, TYPE, IMPL);\
+        try{\
+            auto status = ((FUN)(__VA_ARGS__));\
+            if(ERROR_COND){\
+                ::cv::instr::getCurrentNode()->m_payload.m_funError = true;\
+                CV_INSTRUMENT_MARK_META(IMPL, #FUN " - BadExit");\
+            }\
+            return status;\
+        }catch(...){\
+            ::cv::instr::getCurrentNode()->m_payload.m_funError = true;\
+            CV_INSTRUMENT_MARK_META(IMPL, #FUN " - BadExit");\
+            throw;\
+        }\
+    }else{\
+        return ((FUN)(__VA_ARGS__));\
+    }\
+}())
+// Instrument functions with void return type
+#define CV_INSTRUMENT_FUN_RV_META(TYPE, IMPL, FUN, ...) ([&]()\
+{\
+    if(::cv::instr::useInstrumentation()){\
+        ::cv::instr::IntrumentationRegion __instr__(#FUN, __FILE__, __LINE__, NULL, false, TYPE, IMPL);\
+        try{\
+            (FUN)(__VA_ARGS__);\
+        }catch(...){\
+            ::cv::instr::getCurrentNode()->m_payload.m_funError = true;\
+            CV_INSTRUMENT_MARK_META(IMPL, #FUN "- BadExit");\
+            throw;\
+        }\
+    }else{\
+        (FUN)(__VA_ARGS__);\
+    }\
+}())
+// Instrumentation information marker
+#define CV_INSTRUMENT_MARK_META(IMPL, NAME, ...) {::cv::instr::IntrumentationRegion __instr_mark__(NAME, __FILE__, __LINE__, NULL, false, ::cv::instr::TYPE_MARKER, IMPL);}
+
+///// General instrumentation
+// General OpenCV region instrumentation macro
+#define CV_INSTRUMENT_REGION()              CV_INSTRUMENT_REGION_META(__FUNCTION__, false, ::cv::instr::TYPE_GENERAL, ::cv::instr::IMPL_PLAIN)
+// Custom OpenCV region instrumentation macro
+#define CV_INSTRUMENT_REGION_NAME(NAME)     CV_INSTRUMENT_REGION_CUSTOM_META(NAME,  false, ::cv::instr::TYPE_GENERAL, ::cv::instr::IMPL_PLAIN)
+// Instrumentation for parallel_for_ or other regions which forks and gathers threads
+#define CV_INSTRUMENT_REGION_MT_FORK()      CV_INSTRUMENT_REGION_META(__FUNCTION__, true,  ::cv::instr::TYPE_GENERAL, ::cv::instr::IMPL_PLAIN);
+
+///// IPP instrumentation
+// Wrapper region instrumentation macro
+#define CV_INSTRUMENT_REGION_IPP()          CV_INSTRUMENT_REGION_META(__FUNCTION__, false, ::cv::instr::TYPE_WRAPPER, ::cv::instr::IMPL_IPP)
+// Function instrumentation macro
+#define CV_INSTRUMENT_FUN_IPP(FUN, ...)     CV_INSTRUMENT_FUN_RT_META(::cv::instr::TYPE_FUN, ::cv::instr::IMPL_IPP, status < 0, FUN, __VA_ARGS__)
+// Diagnostic markers
+#define CV_INSTRUMENT_MARK_IPP(NAME)        CV_INSTRUMENT_MARK_META(::cv::instr::IMPL_IPP, NAME)
+
+///// OpenCL instrumentation
+// Wrapper region instrumentation macro
+#define CV_INSTRUMENT_REGION_OPENCL()              CV_INSTRUMENT_REGION_META(__FUNCTION__, false, ::cv::instr::TYPE_WRAPPER, ::cv::instr::IMPL_OPENCL)
+// OpenCL kernel compilation wrapper
+#define CV_INSTRUMENT_REGION_OPENCL_COMPILE(NAME)  CV_INSTRUMENT_REGION_META(NAME, false, ::cv::instr::TYPE_WRAPPER, ::cv::instr::IMPL_OPENCL)
+// OpenCL kernel run wrapper
+#define CV_INSTRUMENT_REGION_OPENCL_RUN(NAME)      CV_INSTRUMENT_REGION_META(NAME, false, ::cv::instr::TYPE_FUN, ::cv::instr::IMPL_OPENCL)
+// Diagnostic markers
+#define CV_INSTRUMENT_MARK_OPENCL(NAME)            CV_INSTRUMENT_MARK_META(::cv::instr::IMPL_OPENCL, NAME)
+#else
+#define CV_INSTRUMENT_REGION_META(...)
+
+#define CV_INSTRUMENT_REGION()
+#define CV_INSTRUMENT_REGION_NAME(...)
+#define CV_INSTRUMENT_REGION_MT_FORK()
+
+#define CV_INSTRUMENT_REGION_IPP()
+#define CV_INSTRUMENT_FUN_IPP(FUN, ...) ((FUN)(__VA_ARGS__))
+#define CV_INSTRUMENT_MARK_IPP(...)
+
+#define CV_INSTRUMENT_REGION_OPENCL()
+#define CV_INSTRUMENT_REGION_OPENCL_COMPILE(...)
+#define CV_INSTRUMENT_REGION_OPENCL_RUN(...)
+#define CV_INSTRUMENT_MARK_OPENCL(...)
+#endif
+
+//! @endcond
+
+#endif // OPENCV_CORE_PRIVATE_HPP

include/opencv2/core/ptr.inl.hpp

@@ -0,0 +1,379 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                           License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2013, NVIDIA Corporation, all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the copyright holders or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef OPENCV_CORE_PTR_INL_HPP
+#define OPENCV_CORE_PTR_INL_HPP
+
+#include <algorithm>
+
+//! @cond IGNORED
+
+namespace cv {
+
+template<typename Y>
+void DefaultDeleter<Y>::operator () (Y* p) const
+{
+    delete p;
+}
+
+namespace detail
+{
+
+struct PtrOwner
+{
+    PtrOwner() : refCount(1)
+    {}
+
+    void incRef()
+    {
+        CV_XADD(&refCount, 1);
+    }
+
+    void decRef()
+    {
+        if (CV_XADD(&refCount, -1) == 1) deleteSelf();
+    }
+
+protected:
+    /* This doesn't really need to be virtual, since PtrOwner is never deleted
+       directly, but it doesn't hurt and it helps avoid warnings. */
+    virtual ~PtrOwner()
+    {}
+
+    virtual void deleteSelf() = 0;
+
+private:
+    unsigned int refCount;
+
+    // noncopyable
+    PtrOwner(const PtrOwner&);
+    PtrOwner& operator = (const PtrOwner&);
+};
+
+template<typename Y, typename D>
+struct PtrOwnerImpl : PtrOwner
+{
+    PtrOwnerImpl(Y* p, D d) : owned(p), deleter(d)
+    {}
+
+    void deleteSelf()
+    {
+        deleter(owned);
+        delete this;
+    }
+
+private:
+    Y* owned;
+    D deleter;
+};
+
+
+}
+
+template<typename T>
+Ptr<T>::Ptr() : owner(NULL), stored(NULL)
+{}
+
+template<typename T>
+template<typename Y>
+Ptr<T>::Ptr(Y* p)
+  : owner(p
+      ? new detail::PtrOwnerImpl<Y, DefaultDeleter<Y> >(p, DefaultDeleter<Y>())
+      : NULL),
+    stored(p)
+{}
+
+template<typename T>
+template<typename Y, typename D>
+Ptr<T>::Ptr(Y* p, D d)
+  : owner(p
+      ? new detail::PtrOwnerImpl<Y, D>(p, d)
+      : NULL),
+    stored(p)
+{}
+
+template<typename T>
+Ptr<T>::Ptr(const Ptr& o) : owner(o.owner), stored(o.stored)
+{
+    if (owner) owner->incRef();
+}
+
+template<typename T>
+template<typename Y>
+Ptr<T>::Ptr(const Ptr<Y>& o) : owner(o.owner), stored(o.stored)
+{
+    if (owner) owner->incRef();
+}
+
+template<typename T>
+template<typename Y>
+Ptr<T>::Ptr(const Ptr<Y>& o, T* p) : owner(o.owner), stored(p)
+{
+    if (owner) owner->incRef();
+}
+
+template<typename T>
+Ptr<T>::~Ptr()
+{
+    release();
+}
+
+template<typename T>
+Ptr<T>& Ptr<T>::operator = (const Ptr<T>& o)
+{
+    Ptr(o).swap(*this);
+    return *this;
+}
+
+template<typename T>
+template<typename Y>
+Ptr<T>& Ptr<T>::operator = (const Ptr<Y>& o)
+{
+    Ptr(o).swap(*this);
+    return *this;
+}
+
+template<typename T>
+void Ptr<T>::release()
+{
+    if (owner) owner->decRef();
+    owner = NULL;
+    stored = NULL;
+}
+
+template<typename T>
+template<typename Y>
+void Ptr<T>::reset(Y* p)
+{
+    Ptr(p).swap(*this);
+}
+
+template<typename T>
+template<typename Y, typename D>
+void Ptr<T>::reset(Y* p, D d)
+{
+    Ptr(p, d).swap(*this);
+}
+
+template<typename T>
+void Ptr<T>::swap(Ptr<T>& o)
+{
+    std::swap(owner, o.owner);
+    std::swap(stored, o.stored);
+}
+
+template<typename T>
+T* Ptr<T>::get() const
+{
+    return stored;
+}
+
+template<typename T>
+typename detail::RefOrVoid<T>::type Ptr<T>::operator * () const
+{
+    return *stored;
+}
+
+template<typename T>
+T* Ptr<T>::operator -> () const
+{
+    return stored;
+}
+
+template<typename T>
+Ptr<T>::operator T* () const
+{
+    return stored;
+}
+
+
+template<typename T>
+bool Ptr<T>::empty() const
+{
+    return !stored;
+}
+
+template<typename T>
+template<typename Y>
+Ptr<Y> Ptr<T>::staticCast() const
+{
+    return Ptr<Y>(*this, static_cast<Y*>(stored));
+}
+
+template<typename T>
+template<typename Y>
+Ptr<Y> Ptr<T>::constCast() const
+{
+    return Ptr<Y>(*this, const_cast<Y*>(stored));
+}
+
+template<typename T>
+template<typename Y>
+Ptr<Y> Ptr<T>::dynamicCast() const
+{
+    return Ptr<Y>(*this, dynamic_cast<Y*>(stored));
+}
+
+#ifdef CV_CXX_MOVE_SEMANTICS
+
+template<typename T>
+Ptr<T>::Ptr(Ptr&& o) : owner(o.owner), stored(o.stored)
+{
+    o.owner = NULL;
+    o.stored = NULL;
+}
+
+template<typename T>
+Ptr<T>& Ptr<T>::operator = (Ptr<T>&& o)
+{
+    if (this == &o)
+        return *this;
+
+    release();
+    owner = o.owner;
+    stored = o.stored;
+    o.owner = NULL;
+    o.stored = NULL;
+    return *this;
+}
+
+#endif
+
+
+template<typename T>
+void swap(Ptr<T>& ptr1, Ptr<T>& ptr2){
+    ptr1.swap(ptr2);
+}
+
+template<typename T>
+bool operator == (const Ptr<T>& ptr1, const Ptr<T>& ptr2)
+{
+    return ptr1.get() == ptr2.get();
+}
+
+template<typename T>
+bool operator != (const Ptr<T>& ptr1, const Ptr<T>& ptr2)
+{
+    return ptr1.get() != ptr2.get();
+}
+
+template<typename T>
+Ptr<T> makePtr()
+{
+    return Ptr<T>(new T());
+}
+
+template<typename T, typename A1>
+Ptr<T> makePtr(const A1& a1)
+{
+    return Ptr<T>(new T(a1));
+}
+
+template<typename T, typename A1, typename A2>
+Ptr<T> makePtr(const A1& a1, const A2& a2)
+{
+    return Ptr<T>(new T(a1, a2));
+}
+
+template<typename T, typename A1, typename A2, typename A3>
+Ptr<T> makePtr(const A1& a1, const A2& a2, const A3& a3)
+{
+    return Ptr<T>(new T(a1, a2, a3));
+}
+
+template<typename T, typename A1, typename A2, typename A3, typename A4>
+Ptr<T> makePtr(const A1& a1, const A2& a2, const A3& a3, const A4& a4)
+{
+    return Ptr<T>(new T(a1, a2, a3, a4));
+}
+
+template<typename T, typename A1, typename A2, typename A3, typename A4, typename A5>
+Ptr<T> makePtr(const A1& a1, const A2& a2, const A3& a3, const A4& a4, const A5& a5)
+{
+    return Ptr<T>(new T(a1, a2, a3, a4, a5));
+}
+
+template<typename T, typename A1, typename A2, typename A3, typename A4, typename A5, typename A6>
+Ptr<T> makePtr(const A1& a1, const A2& a2, const A3& a3, const A4& a4, const A5& a5, const A6& a6)
+{
+    return Ptr<T>(new T(a1, a2, a3, a4, a5, a6));
+}
+
+template<typename T, typename A1, typename A2, typename A3, typename A4, typename A5, typename A6, typename A7>
+Ptr<T> makePtr(const A1& a1, const A2& a2, const A3& a3, const A4& a4, const A5& a5, const A6& a6, const A7& a7)
+{
+    return Ptr<T>(new T(a1, a2, a3, a4, a5, a6, a7));
+}
+
+template<typename T, typename A1, typename A2, typename A3, typename A4, typename A5, typename A6, typename A7, typename A8>
+Ptr<T> makePtr(const A1& a1, const A2& a2, const A3& a3, const A4& a4, const A5& a5, const A6& a6, const A7& a7, const A8& a8)
+{
+    return Ptr<T>(new T(a1, a2, a3, a4, a5, a6, a7, a8));
+}
+
+template<typename T, typename A1, typename A2, typename A3, typename A4, typename A5, typename A6, typename A7, typename A8, typename A9>
+Ptr<T> makePtr(const A1& a1, const A2& a2, const A3& a3, const A4& a4, const A5& a5, const A6& a6, const A7& a7, const A8& a8, const A9& a9)
+{
+    return Ptr<T>(new T(a1, a2, a3, a4, a5, a6, a7, a8, a9));
+}
+
+template<typename T, typename A1, typename A2, typename A3, typename A4, typename A5, typename A6, typename A7, typename A8, typename A9, typename A10>
+Ptr<T> makePtr(const A1& a1, const A2& a2, const A3& a3, const A4& a4, const A5& a5, const A6& a6, const A7& a7, const A8& a8, const A9& a9, const A10& a10)
+{
+    return Ptr<T>(new T(a1, a2, a3, a4, a5, a6, a7, a8, a9, a10));
+}
+
+template<typename T, typename A1, typename A2, typename A3, typename A4, typename A5, typename A6, typename A7, typename A8, typename A9, typename A10, typename A11>
+Ptr<T> makePtr(const A1& a1, const A2& a2, const A3& a3, const A4& a4, const A5& a5, const A6& a6, const A7& a7, const A8& a8, const A9& a9, const A10& a10, const A11& a11)
+{
+    return Ptr<T>(new T(a1, a2, a3, a4, a5, a6, a7, a8, a9, a10, a11));
+}
+
+template<typename T, typename A1, typename A2, typename A3, typename A4, typename A5, typename A6, typename A7, typename A8, typename A9, typename A10, typename A11, typename A12>
+Ptr<T> makePtr(const A1& a1, const A2& a2, const A3& a3, const A4& a4, const A5& a5, const A6& a6, const A7& a7, const A8& a8, const A9& a9, const A10& a10, const A11& a11, const A12& a12)
+{
+    return Ptr<T>(new T(a1, a2, a3, a4, a5, a6, a7, a8, a9, a10, a11, a12));
+}
+} // namespace cv
+
+//! @endcond
+
+#endif // OPENCV_CORE_PTR_INL_HPP

include/opencv2/core/saturate.hpp

@@ -0,0 +1,150 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                          License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
+// Copyright (C) 2014, Itseez Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef OPENCV_CORE_SATURATE_HPP
+#define OPENCV_CORE_SATURATE_HPP
+
+#include "opencv2/core/cvdef.h"
+#include "opencv2/core/fast_math.hpp"
+
+namespace cv
+{
+
+//! @addtogroup core_utils
+//! @{
+
+/////////////// saturate_cast (used in image & signal processing) ///////////////////
+
+/** @brief Template function for accurate conversion from one primitive type to another.
+
+ The functions saturate_cast resemble the standard C++ cast operations, such as static_cast\<T\>()
+ and others. They perform an efficient and accurate conversion from one primitive type to another
+ (see the introduction chapter). saturate in the name means that when the input value v is out of the
+ range of the target type, the result is not formed just by taking low bits of the input, but instead
+ the value is clipped. For example:
+ @code
+ uchar a = saturate_cast<uchar>(-100); // a = 0 (UCHAR_MIN)
+ short b = saturate_cast<short>(33333.33333); // b = 32767 (SHRT_MAX)
+ @endcode
+ Such clipping is done when the target type is unsigned char , signed char , unsigned short or
+ signed short . For 32-bit integers, no clipping is done.
+
+ When the parameter is a floating-point value and the target type is an integer (8-, 16- or 32-bit),
+ the floating-point value is first rounded to the nearest integer and then clipped if needed (when
+ the target type is 8- or 16-bit).
+
+ This operation is used in the simplest or most complex image processing functions in OpenCV.
+
+ @param v Function parameter.
+ @sa add, subtract, multiply, divide, Mat::convertTo
+ */
+template<typename _Tp> static inline _Tp saturate_cast(uchar v)    { return _Tp(v); }
+/** @overload */
+template<typename _Tp> static inline _Tp saturate_cast(schar v)    { return _Tp(v); }
+/** @overload */
+template<typename _Tp> static inline _Tp saturate_cast(ushort v)   { return _Tp(v); }
+/** @overload */
+template<typename _Tp> static inline _Tp saturate_cast(short v)    { return _Tp(v); }
+/** @overload */
+template<typename _Tp> static inline _Tp saturate_cast(unsigned v) { return _Tp(v); }
+/** @overload */
+template<typename _Tp> static inline _Tp saturate_cast(int v)      { return _Tp(v); }
+/** @overload */
+template<typename _Tp> static inline _Tp saturate_cast(float v)    { return _Tp(v); }
+/** @overload */
+template<typename _Tp> static inline _Tp saturate_cast(double v)   { return _Tp(v); }
+/** @overload */
+template<typename _Tp> static inline _Tp saturate_cast(int64 v)    { return _Tp(v); }
+/** @overload */
+template<typename _Tp> static inline _Tp saturate_cast(uint64 v)   { return _Tp(v); }
+
+template<> inline uchar saturate_cast<uchar>(schar v)        { return (uchar)std::max((int)v, 0); }
+template<> inline uchar saturate_cast<uchar>(ushort v)       { return (uchar)std::min((unsigned)v, (unsigned)UCHAR_MAX); }
+template<> inline uchar saturate_cast<uchar>(int v)          { return (uchar)((unsigned)v <= UCHAR_MAX ? v : v > 0 ? UCHAR_MAX : 0); }
+template<> inline uchar saturate_cast<uchar>(short v)        { return saturate_cast<uchar>((int)v); }
+template<> inline uchar saturate_cast<uchar>(unsigned v)     { return (uchar)std::min(v, (unsigned)UCHAR_MAX); }
+template<> inline uchar saturate_cast<uchar>(float v)        { int iv = cvRound(v); return saturate_cast<uchar>(iv); }
+template<> inline uchar saturate_cast<uchar>(double v)       { int iv = cvRound(v); return saturate_cast<uchar>(iv); }
+template<> inline uchar saturate_cast<uchar>(int64 v)        { return (uchar)((uint64)v <= (uint64)UCHAR_MAX ? v : v > 0 ? UCHAR_MAX : 0); }
+template<> inline uchar saturate_cast<uchar>(uint64 v)       { return (uchar)std::min(v, (uint64)UCHAR_MAX); }
+
+template<> inline schar saturate_cast<schar>(uchar v)        { return (schar)std::min((int)v, SCHAR_MAX); }
+template<> inline schar saturate_cast<schar>(ushort v)       { return (schar)std::min((unsigned)v, (unsigned)SCHAR_MAX); }
+template<> inline schar saturate_cast<schar>(int v)          { return (schar)((unsigned)(v-SCHAR_MIN) <= (unsigned)UCHAR_MAX ? v : v > 0 ? SCHAR_MAX : SCHAR_MIN); }
+template<> inline schar saturate_cast<schar>(short v)        { return saturate_cast<schar>((int)v); }
+template<> inline schar saturate_cast<schar>(unsigned v)     { return (schar)std::min(v, (unsigned)SCHAR_MAX); }
+template<> inline schar saturate_cast<schar>(float v)        { int iv = cvRound(v); return saturate_cast<schar>(iv); }
+template<> inline schar saturate_cast<schar>(double v)       { int iv = cvRound(v); return saturate_cast<schar>(iv); }
+template<> inline schar saturate_cast<schar>(int64 v)        { return (schar)((uint64)((int64)v-SCHAR_MIN) <= (uint64)UCHAR_MAX ? v : v > 0 ? SCHAR_MAX : SCHAR_MIN); }
+template<> inline schar saturate_cast<schar>(uint64 v)       { return (schar)std::min(v, (uint64)SCHAR_MAX); }
+
+template<> inline ushort saturate_cast<ushort>(schar v)      { return (ushort)std::max((int)v, 0); }
+template<> inline ushort saturate_cast<ushort>(short v)      { return (ushort)std::max((int)v, 0); }
+template<> inline ushort saturate_cast<ushort>(int v)        { return (ushort)((unsigned)v <= (unsigned)USHRT_MAX ? v : v > 0 ? USHRT_MAX : 0); }
+template<> inline ushort saturate_cast<ushort>(unsigned v)   { return (ushort)std::min(v, (unsigned)USHRT_MAX); }
+template<> inline ushort saturate_cast<ushort>(float v)      { int iv = cvRound(v); return saturate_cast<ushort>(iv); }
+template<> inline ushort saturate_cast<ushort>(double v)     { int iv = cvRound(v); return saturate_cast<ushort>(iv); }
+template<> inline ushort saturate_cast<ushort>(int64 v)      { return (ushort)((uint64)v <= (uint64)USHRT_MAX ? v : v > 0 ? USHRT_MAX : 0); }
+template<> inline ushort saturate_cast<ushort>(uint64 v)     { return (ushort)std::min(v, (uint64)USHRT_MAX); }
+
+template<> inline short saturate_cast<short>(ushort v)       { return (short)std::min((int)v, SHRT_MAX); }
+template<> inline short saturate_cast<short>(int v)          { return (short)((unsigned)(v - SHRT_MIN) <= (unsigned)USHRT_MAX ? v : v > 0 ? SHRT_MAX : SHRT_MIN); }
+template<> inline short saturate_cast<short>(unsigned v)     { return (short)std::min(v, (unsigned)SHRT_MAX); }
+template<> inline short saturate_cast<short>(float v)        { int iv = cvRound(v); return saturate_cast<short>(iv); }
+template<> inline short saturate_cast<short>(double v)       { int iv = cvRound(v); return saturate_cast<short>(iv); }
+template<> inline short saturate_cast<short>(int64 v)        { return (short)((uint64)((int64)v - SHRT_MIN) <= (uint64)USHRT_MAX ? v : v > 0 ? SHRT_MAX : SHRT_MIN); }
+template<> inline short saturate_cast<short>(uint64 v)       { return (short)std::min(v, (uint64)SHRT_MAX); }
+
+template<> inline int saturate_cast<int>(float v)            { return cvRound(v); }
+template<> inline int saturate_cast<int>(double v)           { return cvRound(v); }
+
+// we intentionally do not clip negative numbers, to make -1 become 0xffffffff etc.
+template<> inline unsigned saturate_cast<unsigned>(float v)  { return cvRound(v); }
+template<> inline unsigned saturate_cast<unsigned>(double v) { return cvRound(v); }
+
+//! @}
+
+} // cv
+
+#endif // OPENCV_CORE_SATURATE_HPP

include/opencv2/core/sse_utils.hpp

@@ -0,0 +1,652 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                          License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2015, Itseez Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef OPENCV_CORE_SSE_UTILS_HPP
+#define OPENCV_CORE_SSE_UTILS_HPP
+
+#ifndef __cplusplus
+#  error sse_utils.hpp header must be compiled as C++
+#endif
+
+#include "opencv2/core/cvdef.h"
+
+//! @addtogroup core_utils_sse
+//! @{
+
+#if CV_SSE2
+
+inline void _mm_deinterleave_epi8(__m128i & v_r0, __m128i & v_r1, __m128i & v_g0, __m128i & v_g1)
+{
+    __m128i layer1_chunk0 = _mm_unpacklo_epi8(v_r0, v_g0);
+    __m128i layer1_chunk1 = _mm_unpackhi_epi8(v_r0, v_g0);
+    __m128i layer1_chunk2 = _mm_unpacklo_epi8(v_r1, v_g1);
+    __m128i layer1_chunk3 = _mm_unpackhi_epi8(v_r1, v_g1);
+
+    __m128i layer2_chunk0 = _mm_unpacklo_epi8(layer1_chunk0, layer1_chunk2);
+    __m128i layer2_chunk1 = _mm_unpackhi_epi8(layer1_chunk0, layer1_chunk2);
+    __m128i layer2_chunk2 = _mm_unpacklo_epi8(layer1_chunk1, layer1_chunk3);
+    __m128i layer2_chunk3 = _mm_unpackhi_epi8(layer1_chunk1, layer1_chunk3);
+
+    __m128i layer3_chunk0 = _mm_unpacklo_epi8(layer2_chunk0, layer2_chunk2);
+    __m128i layer3_chunk1 = _mm_unpackhi_epi8(layer2_chunk0, layer2_chunk2);
+    __m128i layer3_chunk2 = _mm_unpacklo_epi8(layer2_chunk1, layer2_chunk3);
+    __m128i layer3_chunk3 = _mm_unpackhi_epi8(layer2_chunk1, layer2_chunk3);
+
+    __m128i layer4_chunk0 = _mm_unpacklo_epi8(layer3_chunk0, layer3_chunk2);
+    __m128i layer4_chunk1 = _mm_unpackhi_epi8(layer3_chunk0, layer3_chunk2);
+    __m128i layer4_chunk2 = _mm_unpacklo_epi8(layer3_chunk1, layer3_chunk3);
+    __m128i layer4_chunk3 = _mm_unpackhi_epi8(layer3_chunk1, layer3_chunk3);
+
+    v_r0 = _mm_unpacklo_epi8(layer4_chunk0, layer4_chunk2);
+    v_r1 = _mm_unpackhi_epi8(layer4_chunk0, layer4_chunk2);
+    v_g0 = _mm_unpacklo_epi8(layer4_chunk1, layer4_chunk3);
+    v_g1 = _mm_unpackhi_epi8(layer4_chunk1, layer4_chunk3);
+}
+
+inline void _mm_deinterleave_epi8(__m128i & v_r0, __m128i & v_r1, __m128i & v_g0,
+                                  __m128i & v_g1, __m128i & v_b0, __m128i & v_b1)
+{
+    __m128i layer1_chunk0 = _mm_unpacklo_epi8(v_r0, v_g1);
+    __m128i layer1_chunk1 = _mm_unpackhi_epi8(v_r0, v_g1);
+    __m128i layer1_chunk2 = _mm_unpacklo_epi8(v_r1, v_b0);
+    __m128i layer1_chunk3 = _mm_unpackhi_epi8(v_r1, v_b0);
+    __m128i layer1_chunk4 = _mm_unpacklo_epi8(v_g0, v_b1);
+    __m128i layer1_chunk5 = _mm_unpackhi_epi8(v_g0, v_b1);
+
+    __m128i layer2_chunk0 = _mm_unpacklo_epi8(layer1_chunk0, layer1_chunk3);
+    __m128i layer2_chunk1 = _mm_unpackhi_epi8(layer1_chunk0, layer1_chunk3);
+    __m128i layer2_chunk2 = _mm_unpacklo_epi8(layer1_chunk1, layer1_chunk4);
+    __m128i layer2_chunk3 = _mm_unpackhi_epi8(layer1_chunk1, layer1_chunk4);
+    __m128i layer2_chunk4 = _mm_unpacklo_epi8(layer1_chunk2, layer1_chunk5);
+    __m128i layer2_chunk5 = _mm_unpackhi_epi8(layer1_chunk2, layer1_chunk5);
+
+    __m128i layer3_chunk0 = _mm_unpacklo_epi8(layer2_chunk0, layer2_chunk3);
+    __m128i layer3_chunk1 = _mm_unpackhi_epi8(layer2_chunk0, layer2_chunk3);
+    __m128i layer3_chunk2 = _mm_unpacklo_epi8(layer2_chunk1, layer2_chunk4);
+    __m128i layer3_chunk3 = _mm_unpackhi_epi8(layer2_chunk1, layer2_chunk4);
+    __m128i layer3_chunk4 = _mm_unpacklo_epi8(layer2_chunk2, layer2_chunk5);
+    __m128i layer3_chunk5 = _mm_unpackhi_epi8(layer2_chunk2, layer2_chunk5);
+
+    __m128i layer4_chunk0 = _mm_unpacklo_epi8(layer3_chunk0, layer3_chunk3);
+    __m128i layer4_chunk1 = _mm_unpackhi_epi8(layer3_chunk0, layer3_chunk3);
+    __m128i layer4_chunk2 = _mm_unpacklo_epi8(layer3_chunk1, layer3_chunk4);
+    __m128i layer4_chunk3 = _mm_unpackhi_epi8(layer3_chunk1, layer3_chunk4);
+    __m128i layer4_chunk4 = _mm_unpacklo_epi8(layer3_chunk2, layer3_chunk5);
+    __m128i layer4_chunk5 = _mm_unpackhi_epi8(layer3_chunk2, layer3_chunk5);
+
+    v_r0 = _mm_unpacklo_epi8(layer4_chunk0, layer4_chunk3);
+    v_r1 = _mm_unpackhi_epi8(layer4_chunk0, layer4_chunk3);
+    v_g0 = _mm_unpacklo_epi8(layer4_chunk1, layer4_chunk4);
+    v_g1 = _mm_unpackhi_epi8(layer4_chunk1, layer4_chunk4);
+    v_b0 = _mm_unpacklo_epi8(layer4_chunk2, layer4_chunk5);
+    v_b1 = _mm_unpackhi_epi8(layer4_chunk2, layer4_chunk5);
+}
+
+inline void _mm_deinterleave_epi8(__m128i & v_r0, __m128i & v_r1, __m128i & v_g0, __m128i & v_g1,
+                                  __m128i & v_b0, __m128i & v_b1, __m128i & v_a0, __m128i & v_a1)
+{
+    __m128i layer1_chunk0 = _mm_unpacklo_epi8(v_r0, v_b0);
+    __m128i layer1_chunk1 = _mm_unpackhi_epi8(v_r0, v_b0);
+    __m128i layer1_chunk2 = _mm_unpacklo_epi8(v_r1, v_b1);
+    __m128i layer1_chunk3 = _mm_unpackhi_epi8(v_r1, v_b1);
+    __m128i layer1_chunk4 = _mm_unpacklo_epi8(v_g0, v_a0);
+    __m128i layer1_chunk5 = _mm_unpackhi_epi8(v_g0, v_a0);
+    __m128i layer1_chunk6 = _mm_unpacklo_epi8(v_g1, v_a1);
+    __m128i layer1_chunk7 = _mm_unpackhi_epi8(v_g1, v_a1);
+
+    __m128i layer2_chunk0 = _mm_unpacklo_epi8(layer1_chunk0, layer1_chunk4);
+    __m128i layer2_chunk1 = _mm_unpackhi_epi8(layer1_chunk0, layer1_chunk4);
+    __m128i layer2_chunk2 = _mm_unpacklo_epi8(layer1_chunk1, layer1_chunk5);
+    __m128i layer2_chunk3 = _mm_unpackhi_epi8(layer1_chunk1, layer1_chunk5);
+    __m128i layer2_chunk4 = _mm_unpacklo_epi8(layer1_chunk2, layer1_chunk6);
+    __m128i layer2_chunk5 = _mm_unpackhi_epi8(layer1_chunk2, layer1_chunk6);
+    __m128i layer2_chunk6 = _mm_unpacklo_epi8(layer1_chunk3, layer1_chunk7);
+    __m128i layer2_chunk7 = _mm_unpackhi_epi8(layer1_chunk3, layer1_chunk7);
+
+    __m128i layer3_chunk0 = _mm_unpacklo_epi8(layer2_chunk0, layer2_chunk4);
+    __m128i layer3_chunk1 = _mm_unpackhi_epi8(layer2_chunk0, layer2_chunk4);
+    __m128i layer3_chunk2 = _mm_unpacklo_epi8(layer2_chunk1, layer2_chunk5);
+    __m128i layer3_chunk3 = _mm_unpackhi_epi8(layer2_chunk1, layer2_chunk5);
+    __m128i layer3_chunk4 = _mm_unpacklo_epi8(layer2_chunk2, layer2_chunk6);
+    __m128i layer3_chunk5 = _mm_unpackhi_epi8(layer2_chunk2, layer2_chunk6);
+    __m128i layer3_chunk6 = _mm_unpacklo_epi8(layer2_chunk3, layer2_chunk7);
+    __m128i layer3_chunk7 = _mm_unpackhi_epi8(layer2_chunk3, layer2_chunk7);
+
+    __m128i layer4_chunk0 = _mm_unpacklo_epi8(layer3_chunk0, layer3_chunk4);
+    __m128i layer4_chunk1 = _mm_unpackhi_epi8(layer3_chunk0, layer3_chunk4);
+    __m128i layer4_chunk2 = _mm_unpacklo_epi8(layer3_chunk1, layer3_chunk5);
+    __m128i layer4_chunk3 = _mm_unpackhi_epi8(layer3_chunk1, layer3_chunk5);
+    __m128i layer4_chunk4 = _mm_unpacklo_epi8(layer3_chunk2, layer3_chunk6);
+    __m128i layer4_chunk5 = _mm_unpackhi_epi8(layer3_chunk2, layer3_chunk6);
+    __m128i layer4_chunk6 = _mm_unpacklo_epi8(layer3_chunk3, layer3_chunk7);
+    __m128i layer4_chunk7 = _mm_unpackhi_epi8(layer3_chunk3, layer3_chunk7);
+
+    v_r0 = _mm_unpacklo_epi8(layer4_chunk0, layer4_chunk4);
+    v_r1 = _mm_unpackhi_epi8(layer4_chunk0, layer4_chunk4);
+    v_g0 = _mm_unpacklo_epi8(layer4_chunk1, layer4_chunk5);
+    v_g1 = _mm_unpackhi_epi8(layer4_chunk1, layer4_chunk5);
+    v_b0 = _mm_unpacklo_epi8(layer4_chunk2, layer4_chunk6);
+    v_b1 = _mm_unpackhi_epi8(layer4_chunk2, layer4_chunk6);
+    v_a0 = _mm_unpacklo_epi8(layer4_chunk3, layer4_chunk7);
+    v_a1 = _mm_unpackhi_epi8(layer4_chunk3, layer4_chunk7);
+}
+
+inline void _mm_interleave_epi8(__m128i & v_r0, __m128i & v_r1, __m128i & v_g0, __m128i & v_g1)
+{
+    __m128i v_mask = _mm_set1_epi16(0x00ff);
+
+    __m128i layer4_chunk0 = _mm_packus_epi16(_mm_and_si128(v_r0, v_mask), _mm_and_si128(v_r1, v_mask));
+    __m128i layer4_chunk2 = _mm_packus_epi16(_mm_srli_epi16(v_r0, 8), _mm_srli_epi16(v_r1, 8));
+    __m128i layer4_chunk1 = _mm_packus_epi16(_mm_and_si128(v_g0, v_mask), _mm_and_si128(v_g1, v_mask));
+    __m128i layer4_chunk3 = _mm_packus_epi16(_mm_srli_epi16(v_g0, 8), _mm_srli_epi16(v_g1, 8));
+
+    __m128i layer3_chunk0 = _mm_packus_epi16(_mm_and_si128(layer4_chunk0, v_mask), _mm_and_si128(layer4_chunk1, v_mask));
+    __m128i layer3_chunk2 = _mm_packus_epi16(_mm_srli_epi16(layer4_chunk0, 8), _mm_srli_epi16(layer4_chunk1, 8));
+    __m128i layer3_chunk1 = _mm_packus_epi16(_mm_and_si128(layer4_chunk2, v_mask), _mm_and_si128(layer4_chunk3, v_mask));
+    __m128i layer3_chunk3 = _mm_packus_epi16(_mm_srli_epi16(layer4_chunk2, 8), _mm_srli_epi16(layer4_chunk3, 8));
+
+    __m128i layer2_chunk0 = _mm_packus_epi16(_mm_and_si128(layer3_chunk0, v_mask), _mm_and_si128(layer3_chunk1, v_mask));
+    __m128i layer2_chunk2 = _mm_packus_epi16(_mm_srli_epi16(layer3_chunk0, 8), _mm_srli_epi16(layer3_chunk1, 8));
+    __m128i layer2_chunk1 = _mm_packus_epi16(_mm_and_si128(layer3_chunk2, v_mask), _mm_and_si128(layer3_chunk3, v_mask));
+    __m128i layer2_chunk3 = _mm_packus_epi16(_mm_srli_epi16(layer3_chunk2, 8), _mm_srli_epi16(layer3_chunk3, 8));
+
+    __m128i layer1_chunk0 = _mm_packus_epi16(_mm_and_si128(layer2_chunk0, v_mask), _mm_and_si128(layer2_chunk1, v_mask));
+    __m128i layer1_chunk2 = _mm_packus_epi16(_mm_srli_epi16(layer2_chunk0, 8), _mm_srli_epi16(layer2_chunk1, 8));
+    __m128i layer1_chunk1 = _mm_packus_epi16(_mm_and_si128(layer2_chunk2, v_mask), _mm_and_si128(layer2_chunk3, v_mask));
+    __m128i layer1_chunk3 = _mm_packus_epi16(_mm_srli_epi16(layer2_chunk2, 8), _mm_srli_epi16(layer2_chunk3, 8));
+
+    v_r0 = _mm_packus_epi16(_mm_and_si128(layer1_chunk0, v_mask), _mm_and_si128(layer1_chunk1, v_mask));
+    v_g0 = _mm_packus_epi16(_mm_srli_epi16(layer1_chunk0, 8), _mm_srli_epi16(layer1_chunk1, 8));
+    v_r1 = _mm_packus_epi16(_mm_and_si128(layer1_chunk2, v_mask), _mm_and_si128(layer1_chunk3, v_mask));
+    v_g1 = _mm_packus_epi16(_mm_srli_epi16(layer1_chunk2, 8), _mm_srli_epi16(layer1_chunk3, 8));
+}
+
+inline void _mm_interleave_epi8(__m128i & v_r0, __m128i & v_r1, __m128i & v_g0,
+                                __m128i & v_g1, __m128i & v_b0, __m128i & v_b1)
+{
+    __m128i v_mask = _mm_set1_epi16(0x00ff);
+
+    __m128i layer4_chunk0 = _mm_packus_epi16(_mm_and_si128(v_r0, v_mask), _mm_and_si128(v_r1, v_mask));
+    __m128i layer4_chunk3 = _mm_packus_epi16(_mm_srli_epi16(v_r0, 8), _mm_srli_epi16(v_r1, 8));
+    __m128i layer4_chunk1 = _mm_packus_epi16(_mm_and_si128(v_g0, v_mask), _mm_and_si128(v_g1, v_mask));
+    __m128i layer4_chunk4 = _mm_packus_epi16(_mm_srli_epi16(v_g0, 8), _mm_srli_epi16(v_g1, 8));
+    __m128i layer4_chunk2 = _mm_packus_epi16(_mm_and_si128(v_b0, v_mask), _mm_and_si128(v_b1, v_mask));
+    __m128i layer4_chunk5 = _mm_packus_epi16(_mm_srli_epi16(v_b0, 8), _mm_srli_epi16(v_b1, 8));
+
+    __m128i layer3_chunk0 = _mm_packus_epi16(_mm_and_si128(layer4_chunk0, v_mask), _mm_and_si128(layer4_chunk1, v_mask));
+    __m128i layer3_chunk3 = _mm_packus_epi16(_mm_srli_epi16(layer4_chunk0, 8), _mm_srli_epi16(layer4_chunk1, 8));
+    __m128i layer3_chunk1 = _mm_packus_epi16(_mm_and_si128(layer4_chunk2, v_mask), _mm_and_si128(layer4_chunk3, v_mask));
+    __m128i layer3_chunk4 = _mm_packus_epi16(_mm_srli_epi16(layer4_chunk2, 8), _mm_srli_epi16(layer4_chunk3, 8));
+    __m128i layer3_chunk2 = _mm_packus_epi16(_mm_and_si128(layer4_chunk4, v_mask), _mm_and_si128(layer4_chunk5, v_mask));
+    __m128i layer3_chunk5 = _mm_packus_epi16(_mm_srli_epi16(layer4_chunk4, 8), _mm_srli_epi16(layer4_chunk5, 8));
+
+    __m128i layer2_chunk0 = _mm_packus_epi16(_mm_and_si128(layer3_chunk0, v_mask), _mm_and_si128(layer3_chunk1, v_mask));
+    __m128i layer2_chunk3 = _mm_packus_epi16(_mm_srli_epi16(layer3_chunk0, 8), _mm_srli_epi16(layer3_chunk1, 8));
+    __m128i layer2_chunk1 = _mm_packus_epi16(_mm_and_si128(layer3_chunk2, v_mask), _mm_and_si128(layer3_chunk3, v_mask));
+    __m128i layer2_chunk4 = _mm_packus_epi16(_mm_srli_epi16(layer3_chunk2, 8), _mm_srli_epi16(layer3_chunk3, 8));
+    __m128i layer2_chunk2 = _mm_packus_epi16(_mm_and_si128(layer3_chunk4, v_mask), _mm_and_si128(layer3_chunk5, v_mask));
+    __m128i layer2_chunk5 = _mm_packus_epi16(_mm_srli_epi16(layer3_chunk4, 8), _mm_srli_epi16(layer3_chunk5, 8));
+
+    __m128i layer1_chunk0 = _mm_packus_epi16(_mm_and_si128(layer2_chunk0, v_mask), _mm_and_si128(layer2_chunk1, v_mask));
+    __m128i layer1_chunk3 = _mm_packus_epi16(_mm_srli_epi16(layer2_chunk0, 8), _mm_srli_epi16(layer2_chunk1, 8));
+    __m128i layer1_chunk1 = _mm_packus_epi16(_mm_and_si128(layer2_chunk2, v_mask), _mm_and_si128(layer2_chunk3, v_mask));
+    __m128i layer1_chunk4 = _mm_packus_epi16(_mm_srli_epi16(layer2_chunk2, 8), _mm_srli_epi16(layer2_chunk3, 8));
+    __m128i layer1_chunk2 = _mm_packus_epi16(_mm_and_si128(layer2_chunk4, v_mask), _mm_and_si128(layer2_chunk5, v_mask));
+    __m128i layer1_chunk5 = _mm_packus_epi16(_mm_srli_epi16(layer2_chunk4, 8), _mm_srli_epi16(layer2_chunk5, 8));
+
+    v_r0 = _mm_packus_epi16(_mm_and_si128(layer1_chunk0, v_mask), _mm_and_si128(layer1_chunk1, v_mask));
+    v_g1 = _mm_packus_epi16(_mm_srli_epi16(layer1_chunk0, 8), _mm_srli_epi16(layer1_chunk1, 8));
+    v_r1 = _mm_packus_epi16(_mm_and_si128(layer1_chunk2, v_mask), _mm_and_si128(layer1_chunk3, v_mask));
+    v_b0 = _mm_packus_epi16(_mm_srli_epi16(layer1_chunk2, 8), _mm_srli_epi16(layer1_chunk3, 8));
+    v_g0 = _mm_packus_epi16(_mm_and_si128(layer1_chunk4, v_mask), _mm_and_si128(layer1_chunk5, v_mask));
+    v_b1 = _mm_packus_epi16(_mm_srli_epi16(layer1_chunk4, 8), _mm_srli_epi16(layer1_chunk5, 8));
+}
+
+inline void _mm_interleave_epi8(__m128i & v_r0, __m128i & v_r1, __m128i & v_g0, __m128i & v_g1,
+                                __m128i & v_b0, __m128i & v_b1, __m128i & v_a0, __m128i & v_a1)
+{
+    __m128i v_mask = _mm_set1_epi16(0x00ff);
+
+    __m128i layer4_chunk0 = _mm_packus_epi16(_mm_and_si128(v_r0, v_mask), _mm_and_si128(v_r1, v_mask));
+    __m128i layer4_chunk4 = _mm_packus_epi16(_mm_srli_epi16(v_r0, 8), _mm_srli_epi16(v_r1, 8));
+    __m128i layer4_chunk1 = _mm_packus_epi16(_mm_and_si128(v_g0, v_mask), _mm_and_si128(v_g1, v_mask));
+    __m128i layer4_chunk5 = _mm_packus_epi16(_mm_srli_epi16(v_g0, 8), _mm_srli_epi16(v_g1, 8));
+    __m128i layer4_chunk2 = _mm_packus_epi16(_mm_and_si128(v_b0, v_mask), _mm_and_si128(v_b1, v_mask));
+    __m128i layer4_chunk6 = _mm_packus_epi16(_mm_srli_epi16(v_b0, 8), _mm_srli_epi16(v_b1, 8));
+    __m128i layer4_chunk3 = _mm_packus_epi16(_mm_and_si128(v_a0, v_mask), _mm_and_si128(v_a1, v_mask));
+    __m128i layer4_chunk7 = _mm_packus_epi16(_mm_srli_epi16(v_a0, 8), _mm_srli_epi16(v_a1, 8));
+
+    __m128i layer3_chunk0 = _mm_packus_epi16(_mm_and_si128(layer4_chunk0, v_mask), _mm_and_si128(layer4_chunk1, v_mask));
+    __m128i layer3_chunk4 = _mm_packus_epi16(_mm_srli_epi16(layer4_chunk0, 8), _mm_srli_epi16(layer4_chunk1, 8));
+    __m128i layer3_chunk1 = _mm_packus_epi16(_mm_and_si128(layer4_chunk2, v_mask), _mm_and_si128(layer4_chunk3, v_mask));
+    __m128i layer3_chunk5 = _mm_packus_epi16(_mm_srli_epi16(layer4_chunk2, 8), _mm_srli_epi16(layer4_chunk3, 8));
+    __m128i layer3_chunk2 = _mm_packus_epi16(_mm_and_si128(layer4_chunk4, v_mask), _mm_and_si128(layer4_chunk5, v_mask));
+    __m128i layer3_chunk6 = _mm_packus_epi16(_mm_srli_epi16(layer4_chunk4, 8), _mm_srli_epi16(layer4_chunk5, 8));
+    __m128i layer3_chunk3 = _mm_packus_epi16(_mm_and_si128(layer4_chunk6, v_mask), _mm_and_si128(layer4_chunk7, v_mask));
+    __m128i layer3_chunk7 = _mm_packus_epi16(_mm_srli_epi16(layer4_chunk6, 8), _mm_srli_epi16(layer4_chunk7, 8));
+
+    __m128i layer2_chunk0 = _mm_packus_epi16(_mm_and_si128(layer3_chunk0, v_mask), _mm_and_si128(layer3_chunk1, v_mask));
+    __m128i layer2_chunk4 = _mm_packus_epi16(_mm_srli_epi16(layer3_chunk0, 8), _mm_srli_epi16(layer3_chunk1, 8));
+    __m128i layer2_chunk1 = _mm_packus_epi16(_mm_and_si128(layer3_chunk2, v_mask), _mm_and_si128(layer3_chunk3, v_mask));
+    __m128i layer2_chunk5 = _mm_packus_epi16(_mm_srli_epi16(layer3_chunk2, 8), _mm_srli_epi16(layer3_chunk3, 8));
+    __m128i layer2_chunk2 = _mm_packus_epi16(_mm_and_si128(layer3_chunk4, v_mask), _mm_and_si128(layer3_chunk5, v_mask));
+    __m128i layer2_chunk6 = _mm_packus_epi16(_mm_srli_epi16(layer3_chunk4, 8), _mm_srli_epi16(layer3_chunk5, 8));
+    __m128i layer2_chunk3 = _mm_packus_epi16(_mm_and_si128(layer3_chunk6, v_mask), _mm_and_si128(layer3_chunk7, v_mask));
+    __m128i layer2_chunk7 = _mm_packus_epi16(_mm_srli_epi16(layer3_chunk6, 8), _mm_srli_epi16(layer3_chunk7, 8));
+
+    __m128i layer1_chunk0 = _mm_packus_epi16(_mm_and_si128(layer2_chunk0, v_mask), _mm_and_si128(layer2_chunk1, v_mask));
+    __m128i layer1_chunk4 = _mm_packus_epi16(_mm_srli_epi16(layer2_chunk0, 8), _mm_srli_epi16(layer2_chunk1, 8));
+    __m128i layer1_chunk1 = _mm_packus_epi16(_mm_and_si128(layer2_chunk2, v_mask), _mm_and_si128(layer2_chunk3, v_mask));
+    __m128i layer1_chunk5 = _mm_packus_epi16(_mm_srli_epi16(layer2_chunk2, 8), _mm_srli_epi16(layer2_chunk3, 8));
+    __m128i layer1_chunk2 = _mm_packus_epi16(_mm_and_si128(layer2_chunk4, v_mask), _mm_and_si128(layer2_chunk5, v_mask));
+    __m128i layer1_chunk6 = _mm_packus_epi16(_mm_srli_epi16(layer2_chunk4, 8), _mm_srli_epi16(layer2_chunk5, 8));
+    __m128i layer1_chunk3 = _mm_packus_epi16(_mm_and_si128(layer2_chunk6, v_mask), _mm_and_si128(layer2_chunk7, v_mask));
+    __m128i layer1_chunk7 = _mm_packus_epi16(_mm_srli_epi16(layer2_chunk6, 8), _mm_srli_epi16(layer2_chunk7, 8));
+
+    v_r0 = _mm_packus_epi16(_mm_and_si128(layer1_chunk0, v_mask), _mm_and_si128(layer1_chunk1, v_mask));
+    v_b0 = _mm_packus_epi16(_mm_srli_epi16(layer1_chunk0, 8), _mm_srli_epi16(layer1_chunk1, 8));
+    v_r1 = _mm_packus_epi16(_mm_and_si128(layer1_chunk2, v_mask), _mm_and_si128(layer1_chunk3, v_mask));
+    v_b1 = _mm_packus_epi16(_mm_srli_epi16(layer1_chunk2, 8), _mm_srli_epi16(layer1_chunk3, 8));
+    v_g0 = _mm_packus_epi16(_mm_and_si128(layer1_chunk4, v_mask), _mm_and_si128(layer1_chunk5, v_mask));
+    v_a0 = _mm_packus_epi16(_mm_srli_epi16(layer1_chunk4, 8), _mm_srli_epi16(layer1_chunk5, 8));
+    v_g1 = _mm_packus_epi16(_mm_and_si128(layer1_chunk6, v_mask), _mm_and_si128(layer1_chunk7, v_mask));
+    v_a1 = _mm_packus_epi16(_mm_srli_epi16(layer1_chunk6, 8), _mm_srli_epi16(layer1_chunk7, 8));
+}
+
+inline void _mm_deinterleave_epi16(__m128i & v_r0, __m128i & v_r1, __m128i & v_g0, __m128i & v_g1)
+{
+    __m128i layer1_chunk0 = _mm_unpacklo_epi16(v_r0, v_g0);
+    __m128i layer1_chunk1 = _mm_unpackhi_epi16(v_r0, v_g0);
+    __m128i layer1_chunk2 = _mm_unpacklo_epi16(v_r1, v_g1);
+    __m128i layer1_chunk3 = _mm_unpackhi_epi16(v_r1, v_g1);
+
+    __m128i layer2_chunk0 = _mm_unpacklo_epi16(layer1_chunk0, layer1_chunk2);
+    __m128i layer2_chunk1 = _mm_unpackhi_epi16(layer1_chunk0, layer1_chunk2);
+    __m128i layer2_chunk2 = _mm_unpacklo_epi16(layer1_chunk1, layer1_chunk3);
+    __m128i layer2_chunk3 = _mm_unpackhi_epi16(layer1_chunk1, layer1_chunk3);
+
+    __m128i layer3_chunk0 = _mm_unpacklo_epi16(layer2_chunk0, layer2_chunk2);
+    __m128i layer3_chunk1 = _mm_unpackhi_epi16(layer2_chunk0, layer2_chunk2);
+    __m128i layer3_chunk2 = _mm_unpacklo_epi16(layer2_chunk1, layer2_chunk3);
+    __m128i layer3_chunk3 = _mm_unpackhi_epi16(layer2_chunk1, layer2_chunk3);
+
+    v_r0 = _mm_unpacklo_epi16(layer3_chunk0, layer3_chunk2);
+    v_r1 = _mm_unpackhi_epi16(layer3_chunk0, layer3_chunk2);
+    v_g0 = _mm_unpacklo_epi16(layer3_chunk1, layer3_chunk3);
+    v_g1 = _mm_unpackhi_epi16(layer3_chunk1, layer3_chunk3);
+}
+
+inline void _mm_deinterleave_epi16(__m128i & v_r0, __m128i & v_r1, __m128i & v_g0,
+                                   __m128i & v_g1, __m128i & v_b0, __m128i & v_b1)
+{
+    __m128i layer1_chunk0 = _mm_unpacklo_epi16(v_r0, v_g1);
+    __m128i layer1_chunk1 = _mm_unpackhi_epi16(v_r0, v_g1);
+    __m128i layer1_chunk2 = _mm_unpacklo_epi16(v_r1, v_b0);
+    __m128i layer1_chunk3 = _mm_unpackhi_epi16(v_r1, v_b0);
+    __m128i layer1_chunk4 = _mm_unpacklo_epi16(v_g0, v_b1);
+    __m128i layer1_chunk5 = _mm_unpackhi_epi16(v_g0, v_b1);
+
+    __m128i layer2_chunk0 = _mm_unpacklo_epi16(layer1_chunk0, layer1_chunk3);
+    __m128i layer2_chunk1 = _mm_unpackhi_epi16(layer1_chunk0, layer1_chunk3);
+    __m128i layer2_chunk2 = _mm_unpacklo_epi16(layer1_chunk1, layer1_chunk4);
+    __m128i layer2_chunk3 = _mm_unpackhi_epi16(layer1_chunk1, layer1_chunk4);
+    __m128i layer2_chunk4 = _mm_unpacklo_epi16(layer1_chunk2, layer1_chunk5);
+    __m128i layer2_chunk5 = _mm_unpackhi_epi16(layer1_chunk2, layer1_chunk5);
+
+    __m128i layer3_chunk0 = _mm_unpacklo_epi16(layer2_chunk0, layer2_chunk3);
+    __m128i layer3_chunk1 = _mm_unpackhi_epi16(layer2_chunk0, layer2_chunk3);
+    __m128i layer3_chunk2 = _mm_unpacklo_epi16(layer2_chunk1, layer2_chunk4);
+    __m128i layer3_chunk3 = _mm_unpackhi_epi16(layer2_chunk1, layer2_chunk4);
+    __m128i layer3_chunk4 = _mm_unpacklo_epi16(layer2_chunk2, layer2_chunk5);
+    __m128i layer3_chunk5 = _mm_unpackhi_epi16(layer2_chunk2, layer2_chunk5);
+
+    v_r0 = _mm_unpacklo_epi16(layer3_chunk0, layer3_chunk3);
+    v_r1 = _mm_unpackhi_epi16(layer3_chunk0, layer3_chunk3);
+    v_g0 = _mm_unpacklo_epi16(layer3_chunk1, layer3_chunk4);
+    v_g1 = _mm_unpackhi_epi16(layer3_chunk1, layer3_chunk4);
+    v_b0 = _mm_unpacklo_epi16(layer3_chunk2, layer3_chunk5);
+    v_b1 = _mm_unpackhi_epi16(layer3_chunk2, layer3_chunk5);
+}
+
+inline void _mm_deinterleave_epi16(__m128i & v_r0, __m128i & v_r1, __m128i & v_g0, __m128i & v_g1,
+                                   __m128i & v_b0, __m128i & v_b1, __m128i & v_a0, __m128i & v_a1)
+{
+    __m128i layer1_chunk0 = _mm_unpacklo_epi16(v_r0, v_b0);
+    __m128i layer1_chunk1 = _mm_unpackhi_epi16(v_r0, v_b0);
+    __m128i layer1_chunk2 = _mm_unpacklo_epi16(v_r1, v_b1);
+    __m128i layer1_chunk3 = _mm_unpackhi_epi16(v_r1, v_b1);
+    __m128i layer1_chunk4 = _mm_unpacklo_epi16(v_g0, v_a0);
+    __m128i layer1_chunk5 = _mm_unpackhi_epi16(v_g0, v_a0);
+    __m128i layer1_chunk6 = _mm_unpacklo_epi16(v_g1, v_a1);
+    __m128i layer1_chunk7 = _mm_unpackhi_epi16(v_g1, v_a1);
+
+    __m128i layer2_chunk0 = _mm_unpacklo_epi16(layer1_chunk0, layer1_chunk4);
+    __m128i layer2_chunk1 = _mm_unpackhi_epi16(layer1_chunk0, layer1_chunk4);
+    __m128i layer2_chunk2 = _mm_unpacklo_epi16(layer1_chunk1, layer1_chunk5);
+    __m128i layer2_chunk3 = _mm_unpackhi_epi16(layer1_chunk1, layer1_chunk5);
+    __m128i layer2_chunk4 = _mm_unpacklo_epi16(layer1_chunk2, layer1_chunk6);
+    __m128i layer2_chunk5 = _mm_unpackhi_epi16(layer1_chunk2, layer1_chunk6);
+    __m128i layer2_chunk6 = _mm_unpacklo_epi16(layer1_chunk3, layer1_chunk7);
+    __m128i layer2_chunk7 = _mm_unpackhi_epi16(layer1_chunk3, layer1_chunk7);
+
+    __m128i layer3_chunk0 = _mm_unpacklo_epi16(layer2_chunk0, layer2_chunk4);
+    __m128i layer3_chunk1 = _mm_unpackhi_epi16(layer2_chunk0, layer2_chunk4);
+    __m128i layer3_chunk2 = _mm_unpacklo_epi16(layer2_chunk1, layer2_chunk5);
+    __m128i layer3_chunk3 = _mm_unpackhi_epi16(layer2_chunk1, layer2_chunk5);
+    __m128i layer3_chunk4 = _mm_unpacklo_epi16(layer2_chunk2, layer2_chunk6);
+    __m128i layer3_chunk5 = _mm_unpackhi_epi16(layer2_chunk2, layer2_chunk6);
+    __m128i layer3_chunk6 = _mm_unpacklo_epi16(layer2_chunk3, layer2_chunk7);
+    __m128i layer3_chunk7 = _mm_unpackhi_epi16(layer2_chunk3, layer2_chunk7);
+
+    v_r0 = _mm_unpacklo_epi16(layer3_chunk0, layer3_chunk4);
+    v_r1 = _mm_unpackhi_epi16(layer3_chunk0, layer3_chunk4);
+    v_g0 = _mm_unpacklo_epi16(layer3_chunk1, layer3_chunk5);
+    v_g1 = _mm_unpackhi_epi16(layer3_chunk1, layer3_chunk5);
+    v_b0 = _mm_unpacklo_epi16(layer3_chunk2, layer3_chunk6);
+    v_b1 = _mm_unpackhi_epi16(layer3_chunk2, layer3_chunk6);
+    v_a0 = _mm_unpacklo_epi16(layer3_chunk3, layer3_chunk7);
+    v_a1 = _mm_unpackhi_epi16(layer3_chunk3, layer3_chunk7);
+}
+
+#if CV_SSE4_1
+
+inline void _mm_interleave_epi16(__m128i & v_r0, __m128i & v_r1, __m128i & v_g0, __m128i & v_g1)
+{
+    __m128i v_mask = _mm_set1_epi32(0x0000ffff);
+
+    __m128i layer3_chunk0 = _mm_packus_epi32(_mm_and_si128(v_r0, v_mask), _mm_and_si128(v_r1, v_mask));
+    __m128i layer3_chunk2 = _mm_packus_epi32(_mm_srli_epi32(v_r0, 16), _mm_srli_epi32(v_r1, 16));
+    __m128i layer3_chunk1 = _mm_packus_epi32(_mm_and_si128(v_g0, v_mask), _mm_and_si128(v_g1, v_mask));
+    __m128i layer3_chunk3 = _mm_packus_epi32(_mm_srli_epi32(v_g0, 16), _mm_srli_epi32(v_g1, 16));
+
+    __m128i layer2_chunk0 = _mm_packus_epi32(_mm_and_si128(layer3_chunk0, v_mask), _mm_and_si128(layer3_chunk1, v_mask));
+    __m128i layer2_chunk2 = _mm_packus_epi32(_mm_srli_epi32(layer3_chunk0, 16), _mm_srli_epi32(layer3_chunk1, 16));
+    __m128i layer2_chunk1 = _mm_packus_epi32(_mm_and_si128(layer3_chunk2, v_mask), _mm_and_si128(layer3_chunk3, v_mask));
+    __m128i layer2_chunk3 = _mm_packus_epi32(_mm_srli_epi32(layer3_chunk2, 16), _mm_srli_epi32(layer3_chunk3, 16));
+
+    __m128i layer1_chunk0 = _mm_packus_epi32(_mm_and_si128(layer2_chunk0, v_mask), _mm_and_si128(layer2_chunk1, v_mask));
+    __m128i layer1_chunk2 = _mm_packus_epi32(_mm_srli_epi32(layer2_chunk0, 16), _mm_srli_epi32(layer2_chunk1, 16));
+    __m128i layer1_chunk1 = _mm_packus_epi32(_mm_and_si128(layer2_chunk2, v_mask), _mm_and_si128(layer2_chunk3, v_mask));
+    __m128i layer1_chunk3 = _mm_packus_epi32(_mm_srli_epi32(layer2_chunk2, 16), _mm_srli_epi32(layer2_chunk3, 16));
+
+    v_r0 = _mm_packus_epi32(_mm_and_si128(layer1_chunk0, v_mask), _mm_and_si128(layer1_chunk1, v_mask));
+    v_g0 = _mm_packus_epi32(_mm_srli_epi32(layer1_chunk0, 16), _mm_srli_epi32(layer1_chunk1, 16));
+    v_r1 = _mm_packus_epi32(_mm_and_si128(layer1_chunk2, v_mask), _mm_and_si128(layer1_chunk3, v_mask));
+    v_g1 = _mm_packus_epi32(_mm_srli_epi32(layer1_chunk2, 16), _mm_srli_epi32(layer1_chunk3, 16));
+}
+
+inline void _mm_interleave_epi16(__m128i & v_r0, __m128i & v_r1, __m128i & v_g0,
+                                 __m128i & v_g1, __m128i & v_b0, __m128i & v_b1)
+{
+    __m128i v_mask = _mm_set1_epi32(0x0000ffff);
+
+    __m128i layer3_chunk0 = _mm_packus_epi32(_mm_and_si128(v_r0, v_mask), _mm_and_si128(v_r1, v_mask));
+    __m128i layer3_chunk3 = _mm_packus_epi32(_mm_srli_epi32(v_r0, 16), _mm_srli_epi32(v_r1, 16));
+    __m128i layer3_chunk1 = _mm_packus_epi32(_mm_and_si128(v_g0, v_mask), _mm_and_si128(v_g1, v_mask));
+    __m128i layer3_chunk4 = _mm_packus_epi32(_mm_srli_epi32(v_g0, 16), _mm_srli_epi32(v_g1, 16));
+    __m128i layer3_chunk2 = _mm_packus_epi32(_mm_and_si128(v_b0, v_mask), _mm_and_si128(v_b1, v_mask));
+    __m128i layer3_chunk5 = _mm_packus_epi32(_mm_srli_epi32(v_b0, 16), _mm_srli_epi32(v_b1, 16));
+
+    __m128i layer2_chunk0 = _mm_packus_epi32(_mm_and_si128(layer3_chunk0, v_mask), _mm_and_si128(layer3_chunk1, v_mask));
+    __m128i layer2_chunk3 = _mm_packus_epi32(_mm_srli_epi32(layer3_chunk0, 16), _mm_srli_epi32(layer3_chunk1, 16));
+    __m128i layer2_chunk1 = _mm_packus_epi32(_mm_and_si128(layer3_chunk2, v_mask), _mm_and_si128(layer3_chunk3, v_mask));
+    __m128i layer2_chunk4 = _mm_packus_epi32(_mm_srli_epi32(layer3_chunk2, 16), _mm_srli_epi32(layer3_chunk3, 16));
+    __m128i layer2_chunk2 = _mm_packus_epi32(_mm_and_si128(layer3_chunk4, v_mask), _mm_and_si128(layer3_chunk5, v_mask));
+    __m128i layer2_chunk5 = _mm_packus_epi32(_mm_srli_epi32(layer3_chunk4, 16), _mm_srli_epi32(layer3_chunk5, 16));
+
+    __m128i layer1_chunk0 = _mm_packus_epi32(_mm_and_si128(layer2_chunk0, v_mask), _mm_and_si128(layer2_chunk1, v_mask));
+    __m128i layer1_chunk3 = _mm_packus_epi32(_mm_srli_epi32(layer2_chunk0, 16), _mm_srli_epi32(layer2_chunk1, 16));
+    __m128i layer1_chunk1 = _mm_packus_epi32(_mm_and_si128(layer2_chunk2, v_mask), _mm_and_si128(layer2_chunk3, v_mask));
+    __m128i layer1_chunk4 = _mm_packus_epi32(_mm_srli_epi32(layer2_chunk2, 16), _mm_srli_epi32(layer2_chunk3, 16));
+    __m128i layer1_chunk2 = _mm_packus_epi32(_mm_and_si128(layer2_chunk4, v_mask), _mm_and_si128(layer2_chunk5, v_mask));
+    __m128i layer1_chunk5 = _mm_packus_epi32(_mm_srli_epi32(layer2_chunk4, 16), _mm_srli_epi32(layer2_chunk5, 16));
+
+    v_r0 = _mm_packus_epi32(_mm_and_si128(layer1_chunk0, v_mask), _mm_and_si128(layer1_chunk1, v_mask));
+    v_g1 = _mm_packus_epi32(_mm_srli_epi32(layer1_chunk0, 16), _mm_srli_epi32(layer1_chunk1, 16));
+    v_r1 = _mm_packus_epi32(_mm_and_si128(layer1_chunk2, v_mask), _mm_and_si128(layer1_chunk3, v_mask));
+    v_b0 = _mm_packus_epi32(_mm_srli_epi32(layer1_chunk2, 16), _mm_srli_epi32(layer1_chunk3, 16));
+    v_g0 = _mm_packus_epi32(_mm_and_si128(layer1_chunk4, v_mask), _mm_and_si128(layer1_chunk5, v_mask));
+    v_b1 = _mm_packus_epi32(_mm_srli_epi32(layer1_chunk4, 16), _mm_srli_epi32(layer1_chunk5, 16));
+}
+
+inline void _mm_interleave_epi16(__m128i & v_r0, __m128i & v_r1, __m128i & v_g0, __m128i & v_g1,
+                                 __m128i & v_b0, __m128i & v_b1, __m128i & v_a0, __m128i & v_a1)
+{
+    __m128i v_mask = _mm_set1_epi32(0x0000ffff);
+
+    __m128i layer3_chunk0 = _mm_packus_epi32(_mm_and_si128(v_r0, v_mask), _mm_and_si128(v_r1, v_mask));
+    __m128i layer3_chunk4 = _mm_packus_epi32(_mm_srli_epi32(v_r0, 16), _mm_srli_epi32(v_r1, 16));
+    __m128i layer3_chunk1 = _mm_packus_epi32(_mm_and_si128(v_g0, v_mask), _mm_and_si128(v_g1, v_mask));
+    __m128i layer3_chunk5 = _mm_packus_epi32(_mm_srli_epi32(v_g0, 16), _mm_srli_epi32(v_g1, 16));
+    __m128i layer3_chunk2 = _mm_packus_epi32(_mm_and_si128(v_b0, v_mask), _mm_and_si128(v_b1, v_mask));
+    __m128i layer3_chunk6 = _mm_packus_epi32(_mm_srli_epi32(v_b0, 16), _mm_srli_epi32(v_b1, 16));
+    __m128i layer3_chunk3 = _mm_packus_epi32(_mm_and_si128(v_a0, v_mask), _mm_and_si128(v_a1, v_mask));
+    __m128i layer3_chunk7 = _mm_packus_epi32(_mm_srli_epi32(v_a0, 16), _mm_srli_epi32(v_a1, 16));
+
+    __m128i layer2_chunk0 = _mm_packus_epi32(_mm_and_si128(layer3_chunk0, v_mask), _mm_and_si128(layer3_chunk1, v_mask));
+    __m128i layer2_chunk4 = _mm_packus_epi32(_mm_srli_epi32(layer3_chunk0, 16), _mm_srli_epi32(layer3_chunk1, 16));
+    __m128i layer2_chunk1 = _mm_packus_epi32(_mm_and_si128(layer3_chunk2, v_mask), _mm_and_si128(layer3_chunk3, v_mask));
+    __m128i layer2_chunk5 = _mm_packus_epi32(_mm_srli_epi32(layer3_chunk2, 16), _mm_srli_epi32(layer3_chunk3, 16));
+    __m128i layer2_chunk2 = _mm_packus_epi32(_mm_and_si128(layer3_chunk4, v_mask), _mm_and_si128(layer3_chunk5, v_mask));
+    __m128i layer2_chunk6 = _mm_packus_epi32(_mm_srli_epi32(layer3_chunk4, 16), _mm_srli_epi32(layer3_chunk5, 16));
+    __m128i layer2_chunk3 = _mm_packus_epi32(_mm_and_si128(layer3_chunk6, v_mask), _mm_and_si128(layer3_chunk7, v_mask));
+    __m128i layer2_chunk7 = _mm_packus_epi32(_mm_srli_epi32(layer3_chunk6, 16), _mm_srli_epi32(layer3_chunk7, 16));
+
+    __m128i layer1_chunk0 = _mm_packus_epi32(_mm_and_si128(layer2_chunk0, v_mask), _mm_and_si128(layer2_chunk1, v_mask));
+    __m128i layer1_chunk4 = _mm_packus_epi32(_mm_srli_epi32(layer2_chunk0, 16), _mm_srli_epi32(layer2_chunk1, 16));
+    __m128i layer1_chunk1 = _mm_packus_epi32(_mm_and_si128(layer2_chunk2, v_mask), _mm_and_si128(layer2_chunk3, v_mask));
+    __m128i layer1_chunk5 = _mm_packus_epi32(_mm_srli_epi32(layer2_chunk2, 16), _mm_srli_epi32(layer2_chunk3, 16));
+    __m128i layer1_chunk2 = _mm_packus_epi32(_mm_and_si128(layer2_chunk4, v_mask), _mm_and_si128(layer2_chunk5, v_mask));
+    __m128i layer1_chunk6 = _mm_packus_epi32(_mm_srli_epi32(layer2_chunk4, 16), _mm_srli_epi32(layer2_chunk5, 16));
+    __m128i layer1_chunk3 = _mm_packus_epi32(_mm_and_si128(layer2_chunk6, v_mask), _mm_and_si128(layer2_chunk7, v_mask));
+    __m128i layer1_chunk7 = _mm_packus_epi32(_mm_srli_epi32(layer2_chunk6, 16), _mm_srli_epi32(layer2_chunk7, 16));
+
+    v_r0 = _mm_packus_epi32(_mm_and_si128(layer1_chunk0, v_mask), _mm_and_si128(layer1_chunk1, v_mask));
+    v_b0 = _mm_packus_epi32(_mm_srli_epi32(layer1_chunk0, 16), _mm_srli_epi32(layer1_chunk1, 16));
+    v_r1 = _mm_packus_epi32(_mm_and_si128(layer1_chunk2, v_mask), _mm_and_si128(layer1_chunk3, v_mask));
+    v_b1 = _mm_packus_epi32(_mm_srli_epi32(layer1_chunk2, 16), _mm_srli_epi32(layer1_chunk3, 16));
+    v_g0 = _mm_packus_epi32(_mm_and_si128(layer1_chunk4, v_mask), _mm_and_si128(layer1_chunk5, v_mask));
+    v_a0 = _mm_packus_epi32(_mm_srli_epi32(layer1_chunk4, 16), _mm_srli_epi32(layer1_chunk5, 16));
+    v_g1 = _mm_packus_epi32(_mm_and_si128(layer1_chunk6, v_mask), _mm_and_si128(layer1_chunk7, v_mask));
+    v_a1 = _mm_packus_epi32(_mm_srli_epi32(layer1_chunk6, 16), _mm_srli_epi32(layer1_chunk7, 16));
+}
+
+#endif // CV_SSE4_1
+
+inline void _mm_deinterleave_ps(__m128 & v_r0, __m128 & v_r1, __m128 & v_g0, __m128 & v_g1)
+{
+    __m128 layer1_chunk0 = _mm_unpacklo_ps(v_r0, v_g0);
+    __m128 layer1_chunk1 = _mm_unpackhi_ps(v_r0, v_g0);
+    __m128 layer1_chunk2 = _mm_unpacklo_ps(v_r1, v_g1);
+    __m128 layer1_chunk3 = _mm_unpackhi_ps(v_r1, v_g1);
+
+    __m128 layer2_chunk0 = _mm_unpacklo_ps(layer1_chunk0, layer1_chunk2);
+    __m128 layer2_chunk1 = _mm_unpackhi_ps(layer1_chunk0, layer1_chunk2);
+    __m128 layer2_chunk2 = _mm_unpacklo_ps(layer1_chunk1, layer1_chunk3);
+    __m128 layer2_chunk3 = _mm_unpackhi_ps(layer1_chunk1, layer1_chunk3);
+
+    v_r0 = _mm_unpacklo_ps(layer2_chunk0, layer2_chunk2);
+    v_r1 = _mm_unpackhi_ps(layer2_chunk0, layer2_chunk2);
+    v_g0 = _mm_unpacklo_ps(layer2_chunk1, layer2_chunk3);
+    v_g1 = _mm_unpackhi_ps(layer2_chunk1, layer2_chunk3);
+}
+
+inline void _mm_deinterleave_ps(__m128 & v_r0, __m128 & v_r1, __m128 & v_g0,
+                                __m128 & v_g1, __m128 & v_b0, __m128 & v_b1)
+{
+    __m128 layer1_chunk0 = _mm_unpacklo_ps(v_r0, v_g1);
+    __m128 layer1_chunk1 = _mm_unpackhi_ps(v_r0, v_g1);
+    __m128 layer1_chunk2 = _mm_unpacklo_ps(v_r1, v_b0);
+    __m128 layer1_chunk3 = _mm_unpackhi_ps(v_r1, v_b0);
+    __m128 layer1_chunk4 = _mm_unpacklo_ps(v_g0, v_b1);
+    __m128 layer1_chunk5 = _mm_unpackhi_ps(v_g0, v_b1);
+
+    __m128 layer2_chunk0 = _mm_unpacklo_ps(layer1_chunk0, layer1_chunk3);
+    __m128 layer2_chunk1 = _mm_unpackhi_ps(layer1_chunk0, layer1_chunk3);
+    __m128 layer2_chunk2 = _mm_unpacklo_ps(layer1_chunk1, layer1_chunk4);
+    __m128 layer2_chunk3 = _mm_unpackhi_ps(layer1_chunk1, layer1_chunk4);
+    __m128 layer2_chunk4 = _mm_unpacklo_ps(layer1_chunk2, layer1_chunk5);
+    __m128 layer2_chunk5 = _mm_unpackhi_ps(layer1_chunk2, layer1_chunk5);
+
+    v_r0 = _mm_unpacklo_ps(layer2_chunk0, layer2_chunk3);
+    v_r1 = _mm_unpackhi_ps(layer2_chunk0, layer2_chunk3);
+    v_g0 = _mm_unpacklo_ps(layer2_chunk1, layer2_chunk4);
+    v_g1 = _mm_unpackhi_ps(layer2_chunk1, layer2_chunk4);
+    v_b0 = _mm_unpacklo_ps(layer2_chunk2, layer2_chunk5);
+    v_b1 = _mm_unpackhi_ps(layer2_chunk2, layer2_chunk5);
+}
+
+inline void _mm_deinterleave_ps(__m128 & v_r0, __m128 & v_r1, __m128 & v_g0, __m128 & v_g1,
+                                __m128 & v_b0, __m128 & v_b1, __m128 & v_a0, __m128 & v_a1)
+{
+    __m128 layer1_chunk0 = _mm_unpacklo_ps(v_r0, v_b0);
+    __m128 layer1_chunk1 = _mm_unpackhi_ps(v_r0, v_b0);
+    __m128 layer1_chunk2 = _mm_unpacklo_ps(v_r1, v_b1);
+    __m128 layer1_chunk3 = _mm_unpackhi_ps(v_r1, v_b1);
+    __m128 layer1_chunk4 = _mm_unpacklo_ps(v_g0, v_a0);
+    __m128 layer1_chunk5 = _mm_unpackhi_ps(v_g0, v_a0);
+    __m128 layer1_chunk6 = _mm_unpacklo_ps(v_g1, v_a1);
+    __m128 layer1_chunk7 = _mm_unpackhi_ps(v_g1, v_a1);
+
+    __m128 layer2_chunk0 = _mm_unpacklo_ps(layer1_chunk0, layer1_chunk4);
+    __m128 layer2_chunk1 = _mm_unpackhi_ps(layer1_chunk0, layer1_chunk4);
+    __m128 layer2_chunk2 = _mm_unpacklo_ps(layer1_chunk1, layer1_chunk5);
+    __m128 layer2_chunk3 = _mm_unpackhi_ps(layer1_chunk1, layer1_chunk5);
+    __m128 layer2_chunk4 = _mm_unpacklo_ps(layer1_chunk2, layer1_chunk6);
+    __m128 layer2_chunk5 = _mm_unpackhi_ps(layer1_chunk2, layer1_chunk6);
+    __m128 layer2_chunk6 = _mm_unpacklo_ps(layer1_chunk3, layer1_chunk7);
+    __m128 layer2_chunk7 = _mm_unpackhi_ps(layer1_chunk3, layer1_chunk7);
+
+    v_r0 = _mm_unpacklo_ps(layer2_chunk0, layer2_chunk4);
+    v_r1 = _mm_unpackhi_ps(layer2_chunk0, layer2_chunk4);
+    v_g0 = _mm_unpacklo_ps(layer2_chunk1, layer2_chunk5);
+    v_g1 = _mm_unpackhi_ps(layer2_chunk1, layer2_chunk5);
+    v_b0 = _mm_unpacklo_ps(layer2_chunk2, layer2_chunk6);
+    v_b1 = _mm_unpackhi_ps(layer2_chunk2, layer2_chunk6);
+    v_a0 = _mm_unpacklo_ps(layer2_chunk3, layer2_chunk7);
+    v_a1 = _mm_unpackhi_ps(layer2_chunk3, layer2_chunk7);
+}
+
+inline void _mm_interleave_ps(__m128 & v_r0, __m128 & v_r1, __m128 & v_g0, __m128 & v_g1)
+{
+    const int mask_lo = _MM_SHUFFLE(2, 0, 2, 0), mask_hi = _MM_SHUFFLE(3, 1, 3, 1);
+
+    __m128 layer2_chunk0 = _mm_shuffle_ps(v_r0, v_r1, mask_lo);
+    __m128 layer2_chunk2 = _mm_shuffle_ps(v_r0, v_r1, mask_hi);
+    __m128 layer2_chunk1 = _mm_shuffle_ps(v_g0, v_g1, mask_lo);
+    __m128 layer2_chunk3 = _mm_shuffle_ps(v_g0, v_g1, mask_hi);
+
+    __m128 layer1_chunk0 = _mm_shuffle_ps(layer2_chunk0, layer2_chunk1, mask_lo);
+    __m128 layer1_chunk2 = _mm_shuffle_ps(layer2_chunk0, layer2_chunk1, mask_hi);
+    __m128 layer1_chunk1 = _mm_shuffle_ps(layer2_chunk2, layer2_chunk3, mask_lo);
+    __m128 layer1_chunk3 = _mm_shuffle_ps(layer2_chunk2, layer2_chunk3, mask_hi);
+
+    v_r0 = _mm_shuffle_ps(layer1_chunk0, layer1_chunk1, mask_lo);
+    v_g0 = _mm_shuffle_ps(layer1_chunk0, layer1_chunk1, mask_hi);
+    v_r1 = _mm_shuffle_ps(layer1_chunk2, layer1_chunk3, mask_lo);
+    v_g1 = _mm_shuffle_ps(layer1_chunk2, layer1_chunk3, mask_hi);
+}
+
+inline void _mm_interleave_ps(__m128 & v_r0, __m128 & v_r1, __m128 & v_g0,
+                              __m128 & v_g1, __m128 & v_b0, __m128 & v_b1)
+{
+    const int mask_lo = _MM_SHUFFLE(2, 0, 2, 0), mask_hi = _MM_SHUFFLE(3, 1, 3, 1);
+
+    __m128 layer2_chunk0 = _mm_shuffle_ps(v_r0, v_r1, mask_lo);
+    __m128 layer2_chunk3 = _mm_shuffle_ps(v_r0, v_r1, mask_hi);
+    __m128 layer2_chunk1 = _mm_shuffle_ps(v_g0, v_g1, mask_lo);
+    __m128 layer2_chunk4 = _mm_shuffle_ps(v_g0, v_g1, mask_hi);
+    __m128 layer2_chunk2 = _mm_shuffle_ps(v_b0, v_b1, mask_lo);
+    __m128 layer2_chunk5 = _mm_shuffle_ps(v_b0, v_b1, mask_hi);
+
+    __m128 layer1_chunk0 = _mm_shuffle_ps(layer2_chunk0, layer2_chunk1, mask_lo);
+    __m128 layer1_chunk3 = _mm_shuffle_ps(layer2_chunk0, layer2_chunk1, mask_hi);
+    __m128 layer1_chunk1 = _mm_shuffle_ps(layer2_chunk2, layer2_chunk3, mask_lo);
+    __m128 layer1_chunk4 = _mm_shuffle_ps(layer2_chunk2, layer2_chunk3, mask_hi);
+    __m128 layer1_chunk2 = _mm_shuffle_ps(layer2_chunk4, layer2_chunk5, mask_lo);
+    __m128 layer1_chunk5 = _mm_shuffle_ps(layer2_chunk4, layer2_chunk5, mask_hi);
+
+    v_r0 = _mm_shuffle_ps(layer1_chunk0, layer1_chunk1, mask_lo);
+    v_g1 = _mm_shuffle_ps(layer1_chunk0, layer1_chunk1, mask_hi);
+    v_r1 = _mm_shuffle_ps(layer1_chunk2, layer1_chunk3, mask_lo);
+    v_b0 = _mm_shuffle_ps(layer1_chunk2, layer1_chunk3, mask_hi);
+    v_g0 = _mm_shuffle_ps(layer1_chunk4, layer1_chunk5, mask_lo);
+    v_b1 = _mm_shuffle_ps(layer1_chunk4, layer1_chunk5, mask_hi);
+}
+
+inline void _mm_interleave_ps(__m128 & v_r0, __m128 & v_r1, __m128 & v_g0, __m128 & v_g1,
+                              __m128 & v_b0, __m128 & v_b1, __m128 & v_a0, __m128 & v_a1)
+{
+    const int mask_lo = _MM_SHUFFLE(2, 0, 2, 0), mask_hi = _MM_SHUFFLE(3, 1, 3, 1);
+
+    __m128 layer2_chunk0 = _mm_shuffle_ps(v_r0, v_r1, mask_lo);
+    __m128 layer2_chunk4 = _mm_shuffle_ps(v_r0, v_r1, mask_hi);
+    __m128 layer2_chunk1 = _mm_shuffle_ps(v_g0, v_g1, mask_lo);
+    __m128 layer2_chunk5 = _mm_shuffle_ps(v_g0, v_g1, mask_hi);
+    __m128 layer2_chunk2 = _mm_shuffle_ps(v_b0, v_b1, mask_lo);
+    __m128 layer2_chunk6 = _mm_shuffle_ps(v_b0, v_b1, mask_hi);
+    __m128 layer2_chunk3 = _mm_shuffle_ps(v_a0, v_a1, mask_lo);
+    __m128 layer2_chunk7 = _mm_shuffle_ps(v_a0, v_a1, mask_hi);
+
+    __m128 layer1_chunk0 = _mm_shuffle_ps(layer2_chunk0, layer2_chunk1, mask_lo);
+    __m128 layer1_chunk4 = _mm_shuffle_ps(layer2_chunk0, layer2_chunk1, mask_hi);
+    __m128 layer1_chunk1 = _mm_shuffle_ps(layer2_chunk2, layer2_chunk3, mask_lo);
+    __m128 layer1_chunk5 = _mm_shuffle_ps(layer2_chunk2, layer2_chunk3, mask_hi);
+    __m128 layer1_chunk2 = _mm_shuffle_ps(layer2_chunk4, layer2_chunk5, mask_lo);
+    __m128 layer1_chunk6 = _mm_shuffle_ps(layer2_chunk4, layer2_chunk5, mask_hi);
+    __m128 layer1_chunk3 = _mm_shuffle_ps(layer2_chunk6, layer2_chunk7, mask_lo);
+    __m128 layer1_chunk7 = _mm_shuffle_ps(layer2_chunk6, layer2_chunk7, mask_hi);
+
+    v_r0 = _mm_shuffle_ps(layer1_chunk0, layer1_chunk1, mask_lo);
+    v_b0 = _mm_shuffle_ps(layer1_chunk0, layer1_chunk1, mask_hi);
+    v_r1 = _mm_shuffle_ps(layer1_chunk2, layer1_chunk3, mask_lo);
+    v_b1 = _mm_shuffle_ps(layer1_chunk2, layer1_chunk3, mask_hi);
+    v_g0 = _mm_shuffle_ps(layer1_chunk4, layer1_chunk5, mask_lo);
+    v_a0 = _mm_shuffle_ps(layer1_chunk4, layer1_chunk5, mask_hi);
+    v_g1 = _mm_shuffle_ps(layer1_chunk6, layer1_chunk7, mask_lo);
+    v_a1 = _mm_shuffle_ps(layer1_chunk6, layer1_chunk7, mask_hi);
+}
+
+#endif // CV_SSE2
+
+//! @}
+
+#endif //OPENCV_CORE_SSE_UTILS_HPP

include/opencv2/core/traits.hpp

@@ -0,0 +1,326 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                          License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef OPENCV_CORE_TRAITS_HPP
+#define OPENCV_CORE_TRAITS_HPP
+
+#include "opencv2/core/cvdef.h"
+
+namespace cv
+{
+
+//! @addtogroup core_basic
+//! @{
+
+/** @brief Template "trait" class for OpenCV primitive data types.
+
+A primitive OpenCV data type is one of unsigned char, bool, signed char, unsigned short, signed
+short, int, float, double, or a tuple of values of one of these types, where all the values in the
+tuple have the same type. Any primitive type from the list can be defined by an identifier in the
+form CV_\<bit-depth\>{U|S|F}C(\<number_of_channels\>), for example: uchar \~ CV_8UC1, 3-element
+floating-point tuple \~ CV_32FC3, and so on. A universal OpenCV structure that is able to store a
+single instance of such a primitive data type is Vec. Multiple instances of such a type can be
+stored in a std::vector, Mat, Mat_, SparseMat, SparseMat_, or any other container that is able to
+store Vec instances.
+
+The DataType class is basically used to provide a description of such primitive data types without
+adding any fields or methods to the corresponding classes (and it is actually impossible to add
+anything to primitive C/C++ data types). This technique is known in C++ as class traits. It is not
+DataType itself that is used but its specialized versions, such as:
+@code
+    template<> class DataType<uchar>
+    {
+        typedef uchar value_type;
+        typedef int work_type;
+        typedef uchar channel_type;
+        enum { channel_type = CV_8U, channels = 1, fmt='u', type = CV_8U };
+    };
+    ...
+    template<typename _Tp> DataType<std::complex<_Tp> >
+    {
+        typedef std::complex<_Tp> value_type;
+        typedef std::complex<_Tp> work_type;
+        typedef _Tp channel_type;
+        // DataDepth is another helper trait class
+        enum { depth = DataDepth<_Tp>::value, channels=2,
+            fmt=(channels-1)*256+DataDepth<_Tp>::fmt,
+            type=CV_MAKETYPE(depth, channels) };
+    };
+    ...
+@endcode
+The main purpose of this class is to convert compilation-time type information to an
+OpenCV-compatible data type identifier, for example:
+@code
+    // allocates a 30x40 floating-point matrix
+    Mat A(30, 40, DataType<float>::type);
+
+    Mat B = Mat_<std::complex<double> >(3, 3);
+    // the statement below will print 6, 2 , that is depth == CV_64F, channels == 2
+    cout << B.depth() << ", " << B.channels() << endl;
+@endcode
+So, such traits are used to tell OpenCV which data type you are working with, even if such a type is
+not native to OpenCV. For example, the matrix B initialization above is compiled because OpenCV
+defines the proper specialized template class DataType\<complex\<_Tp\> \> . This mechanism is also
+useful (and used in OpenCV this way) for generic algorithms implementations.
+*/
+template<typename _Tp> class DataType
+{
+public:
+    typedef _Tp         value_type;
+    typedef value_type  work_type;
+    typedef value_type  channel_type;
+    typedef value_type  vec_type;
+    enum { generic_type = 1,
+           depth        = -1,
+           channels     = 1,
+           fmt          = 0,
+           type = CV_MAKETYPE(depth, channels)
+         };
+};
+
+template<> class DataType<bool>
+{
+public:
+    typedef bool        value_type;
+    typedef int         work_type;
+    typedef value_type  channel_type;
+    typedef value_type  vec_type;
+    enum { generic_type = 0,
+           depth        = CV_8U,
+           channels     = 1,
+           fmt          = (int)'u',
+           type         = CV_MAKETYPE(depth, channels)
+         };
+};
+
+template<> class DataType<uchar>
+{
+public:
+    typedef uchar       value_type;
+    typedef int         work_type;
+    typedef value_type  channel_type;
+    typedef value_type  vec_type;
+    enum { generic_type = 0,
+           depth        = CV_8U,
+           channels     = 1,
+           fmt          = (int)'u',
+           type         = CV_MAKETYPE(depth, channels)
+         };
+};
+
+template<> class DataType<schar>
+{
+public:
+    typedef schar       value_type;
+    typedef int         work_type;
+    typedef value_type  channel_type;
+    typedef value_type  vec_type;
+    enum { generic_type = 0,
+           depth        = CV_8S,
+           channels     = 1,
+           fmt          = (int)'c',
+           type         = CV_MAKETYPE(depth, channels)
+         };
+};
+
+template<> class DataType<char>
+{
+public:
+    typedef schar       value_type;
+    typedef int         work_type;
+    typedef value_type  channel_type;
+    typedef value_type  vec_type;
+    enum { generic_type = 0,
+           depth        = CV_8S,
+           channels     = 1,
+           fmt          = (int)'c',
+           type         = CV_MAKETYPE(depth, channels)
+         };
+};
+
+template<> class DataType<ushort>
+{
+public:
+    typedef ushort      value_type;
+    typedef int         work_type;
+    typedef value_type  channel_type;
+    typedef value_type  vec_type;
+    enum { generic_type = 0,
+           depth        = CV_16U,
+           channels     = 1,
+           fmt          = (int)'w',
+           type         = CV_MAKETYPE(depth, channels)
+         };
+};
+
+template<> class DataType<short>
+{
+public:
+    typedef short       value_type;
+    typedef int         work_type;
+    typedef value_type  channel_type;
+    typedef value_type  vec_type;
+    enum { generic_type = 0,
+           depth        = CV_16S,
+           channels     = 1,
+           fmt          = (int)'s',
+           type         = CV_MAKETYPE(depth, channels)
+         };
+};
+
+template<> class DataType<int>
+{
+public:
+    typedef int         value_type;
+    typedef value_type  work_type;
+    typedef value_type  channel_type;
+    typedef value_type  vec_type;
+    enum { generic_type = 0,
+           depth        = CV_32S,
+           channels     = 1,
+           fmt          = (int)'i',
+           type         = CV_MAKETYPE(depth, channels)
+         };
+};
+
+template<> class DataType<float>
+{
+public:
+    typedef float       value_type;
+    typedef value_type  work_type;
+    typedef value_type  channel_type;
+    typedef value_type  vec_type;
+    enum { generic_type = 0,
+           depth        = CV_32F,
+           channels     = 1,
+           fmt          = (int)'f',
+           type         = CV_MAKETYPE(depth, channels)
+         };
+};
+
+template<> class DataType<double>
+{
+public:
+    typedef double      value_type;
+    typedef value_type  work_type;
+    typedef value_type  channel_type;
+    typedef value_type  vec_type;
+    enum { generic_type = 0,
+           depth        = CV_64F,
+           channels     = 1,
+           fmt          = (int)'d',
+           type         = CV_MAKETYPE(depth, channels)
+         };
+};
+
+
+/** @brief A helper class for cv::DataType
+
+The class is specialized for each fundamental numerical data type supported by OpenCV. It provides
+DataDepth<T>::value constant.
+*/
+template<typename _Tp> class DataDepth
+{
+public:
+    enum
+    {
+        value = DataType<_Tp>::depth,
+        fmt   = DataType<_Tp>::fmt
+    };
+};
+
+
+
+template<int _depth> class TypeDepth
+{
+    enum { depth = CV_USRTYPE1 };
+    typedef void value_type;
+};
+
+template<> class TypeDepth<CV_8U>
+{
+    enum { depth = CV_8U };
+    typedef uchar value_type;
+};
+
+template<> class TypeDepth<CV_8S>
+{
+    enum { depth = CV_8S };
+    typedef schar value_type;
+};
+
+template<> class TypeDepth<CV_16U>
+{
+    enum { depth = CV_16U };
+    typedef ushort value_type;
+};
+
+template<> class TypeDepth<CV_16S>
+{
+    enum { depth = CV_16S };
+    typedef short value_type;
+};
+
+template<> class TypeDepth<CV_32S>
+{
+    enum { depth = CV_32S };
+    typedef int value_type;
+};
+
+template<> class TypeDepth<CV_32F>
+{
+    enum { depth = CV_32F };
+    typedef float value_type;
+};
+
+template<> class TypeDepth<CV_64F>
+{
+    enum { depth = CV_64F };
+    typedef double value_type;
+};
+
+//! @}
+
+} // cv
+
+#endif // OPENCV_CORE_TRAITS_HPP

include/opencv2/core/va_intel.hpp

@@ -0,0 +1,77 @@
+// This file is part of OpenCV project.
+// It is subject to the license terms in the LICENSE file found in the top-level directory
+// of this distribution and at http://opencv.org/license.html.
+
+// Copyright (C) 2015, Itseez, Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+
+#ifndef OPENCV_CORE_VA_INTEL_HPP
+#define OPENCV_CORE_VA_INTEL_HPP
+
+#ifndef __cplusplus
+#  error va_intel.hpp header must be compiled as C++
+#endif
+
+#include "opencv2/core.hpp"
+#include "ocl.hpp"
+
+#if defined(HAVE_VA)
+# include "va/va.h"
+#else  // HAVE_VA
+# if !defined(_VA_H_)
+    typedef void* VADisplay;
+    typedef unsigned int VASurfaceID;
+# endif // !_VA_H_
+#endif // HAVE_VA
+
+namespace cv { namespace va_intel {
+
+/** @addtogroup core_va_intel
+This section describes Intel VA-API/OpenCL (CL-VA) interoperability.
+
+To enable CL-VA interoperability support, configure OpenCV using CMake with WITH_VA_INTEL=ON . Currently VA-API is
+supported on Linux only. You should also install Intel Media Server Studio (MSS) to use this feature. You may
+have to specify the path(s) to MSS components for cmake in environment variables: VA_INTEL_MSDK_ROOT for Media SDK
+(default is "/opt/intel/mediasdk"), and VA_INTEL_IOCL_ROOT for Intel OpenCL (default is "/opt/intel/opencl").
+
+To use CL-VA interoperability you should first create VADisplay (libva), and then call initializeContextFromVA()
+function to create OpenCL context and set up interoperability.
+*/
+//! @{
+
+/////////////////// CL-VA Interoperability Functions ///////////////////
+
+namespace ocl {
+using namespace cv::ocl;
+
+// TODO static functions in the Context class
+/** @brief Creates OpenCL context from VA.
+@param display    - VADisplay for which CL interop should be established.
+@param tryInterop - try to set up for interoperability, if true; set up for use slow copy if false.
+@return Returns reference to OpenCL Context
+ */
+CV_EXPORTS Context& initializeContextFromVA(VADisplay display, bool tryInterop = true);
+
+} // namespace cv::va_intel::ocl
+
+/** @brief Converts InputArray to VASurfaceID object.
+@param display - VADisplay object.
+@param src     - source InputArray.
+@param surface - destination VASurfaceID object.
+@param size    - size of image represented by VASurfaceID object.
+ */
+CV_EXPORTS void convertToVASurface(VADisplay display, InputArray src, VASurfaceID surface, Size size);
+
+/** @brief Converts VASurfaceID object to OutputArray.
+@param display - VADisplay object.
+@param surface - source VASurfaceID object.
+@param size    - size of image represented by VASurfaceID object.
+@param dst     - destination OutputArray.
+ */
+CV_EXPORTS void convertFromVASurface(VADisplay display, VASurfaceID surface, Size size, OutputArray dst);
+
+//! @}
+
+}} // namespace cv::va_intel
+
+#endif /* OPENCV_CORE_VA_INTEL_HPP */

include/opencv2/core/version.hpp

@@ -10,7 +10,10 @@
 //                        Intel License Agreement
 //                For Open Source Computer Vision Library
 //
-// Copyright( C) 2000, Intel Corporation, all rights reserved.
+// Copyright( C) 2000-2015, Intel Corporation, all rights reserved.
+// Copyright (C) 2011-2013, NVIDIA Corporation, all rights reserved.
+// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
+// Copyright (C) 2015, Itseez Inc., all rights reserved.
 // Third party copyrights are property of their respective owners.
 //
 // Redistribution and use in source and binary forms, with or without modification,
@@ -44,29 +47,25 @@
   Usefull to test in user programs
 */
 
-#ifndef __OPENCV_VERSION_HPP__
-#define __OPENCV_VERSION_HPP__
+#ifndef OPENCV_VERSION_HPP
+#define OPENCV_VERSION_HPP
 
-#define CV_VERSION_EPOCH    2
-#define CV_VERSION_MAJOR    4
-#define CV_VERSION_MINOR    8
+#define CV_VERSION_MAJOR    3
+#define CV_VERSION_MINOR    2
 #define CV_VERSION_REVISION 0
+#define CV_VERSION_STATUS   ""
 
 #define CVAUX_STR_EXP(__A)  #__A
 #define CVAUX_STR(__A)      CVAUX_STR_EXP(__A)
 
-#define CVAUX_STRW_EXP(__A)  L#__A
+#define CVAUX_STRW_EXP(__A)  L ## #__A
 #define CVAUX_STRW(__A)      CVAUX_STRW_EXP(__A)
 
-#if CV_VERSION_REVISION
-#  define CV_VERSION        CVAUX_STR(CV_VERSION_EPOCH) "." CVAUX_STR(CV_VERSION_MAJOR) "." CVAUX_STR(CV_VERSION_MINOR) "." CVAUX_STR(CV_VERSION_REVISION)
-#else
-#  define CV_VERSION        CVAUX_STR(CV_VERSION_EPOCH) "." CVAUX_STR(CV_VERSION_MAJOR) "." CVAUX_STR(CV_VERSION_MINOR)
-#endif
+#define CV_VERSION          CVAUX_STR(CV_VERSION_MAJOR) "." CVAUX_STR(CV_VERSION_MINOR) "." CVAUX_STR(CV_VERSION_REVISION) CV_VERSION_STATUS
 
 /* old  style version constants*/
-#define CV_MAJOR_VERSION    CV_VERSION_EPOCH
-#define CV_MINOR_VERSION    CV_VERSION_MAJOR
-#define CV_SUBMINOR_VERSION CV_VERSION_MINOR
+#define CV_MAJOR_VERSION    CV_VERSION_MAJOR
+#define CV_MINOR_VERSION    CV_VERSION_MINOR
+#define CV_SUBMINOR_VERSION CV_VERSION_REVISION
 
 #endif

include/opencv2/core/wimage.hpp

@@ -1,4 +1,4 @@
-///////////////////////////////////////////////////////////////////////////////
+/*M//////////////////////////////////////////////////////////////////////////////
 //  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
 //
 //  By downloading, copying, installing or using the software you agree to
@@ -36,69 +36,11 @@
 // and on any theory of liability, whether in contract, strict liability,
 // or tort (including negligence or otherwise) arising in any way out of
 // the use of this software, even if advised of the possibility of such damage.
-
-
 /////////////////////////////////////////////////////////////////////////////////
-//
-// Image class which provides a thin layer around an IplImage.  The goals
-// of the class design are:
-//    1. All the data has explicit ownership to avoid memory leaks
-//    2. No hidden allocations or copies for performance.
-//    3. Easy access to OpenCV methods (which will access IPP if available)
-//    4. Can easily treat external data as an image
-//    5. Easy to create images which are subsets of other images
-//    6. Fast pixel access which can take advantage of number of channels
-//          if known at compile time.
-//
-// The WImage class is the image class which provides the data accessors.
-// The 'W' comes from the fact that it is also a wrapper around the popular
-// but inconvenient IplImage class. A WImage can be constructed either using a
-// WImageBuffer class which allocates and frees the data,
-// or using a WImageView class which constructs a subimage or a view into
-// external data.  The view class does no memory management.  Each class
-// actually has two versions, one when the number of channels is known at
-// compile time and one when it isn't.  Using the one with the number of
-// channels specified can provide some compile time optimizations by using the
-// fact that the number of channels is a constant.
-//
-// We use the convention (c,r) to refer to column c and row r with (0,0) being
-// the upper left corner.  This is similar to standard Euclidean coordinates
-// with the first coordinate varying in the horizontal direction and the second
-// coordinate varying in the vertical direction.
-// Thus (c,r) is usually in the domain [0, width) X [0, height)
-//
-// Example usage:
-// WImageBuffer3_b  im(5,7);  // Make a 5X7 3 channel image of type uchar
-// WImageView3_b  sub_im(im, 2,2, 3,3); // 3X3 submatrix
-// vector<float> vec(10, 3.0f);
-// WImageView1_f user_im(&vec[0], 2, 5);  // 2X5 image w/ supplied data
-//
-// im.SetZero();  // same as cvSetZero(im.Ipl())
-// *im(2, 3) = 15;  // Modify the element at column 2, row 3
-// MySetRand(&sub_im);
-//
-// // Copy the second row into the first.  This can be done with no memory
-// // allocation and will use SSE if IPP is available.
-// int w = im.Width();
-// im.View(0,0, w,1).CopyFrom(im.View(0,1, w,1));
-//
-// // Doesn't care about source of data since using WImage
-// void MySetRand(WImage_b* im) { // Works with any number of channels
-//   for (int r = 0; r < im->Height(); ++r) {
-//     float* row = im->Row(r);
-//     for (int c = 0; c < im->Width(); ++c) {
-//        for (int ch = 0; ch < im->Channels(); ++ch, ++row) {
-//          *row = uchar(rand() & 255);
-//        }
-//     }
-//   }
-// }
-//
-// Functions that are not part of the basic image allocation, viewing, and
-// access should come from OpenCV, except some useful functions that are not
-// part of OpenCV can be found in wimage_util.h
-#ifndef __OPENCV_CORE_WIMAGE_HPP__
-#define __OPENCV_CORE_WIMAGE_HPP__
+//M*/
+
+#ifndef OPENCV_CORE_WIMAGE_HPP
+#define OPENCV_CORE_WIMAGE_HPP
 
 #include "opencv2/core/core_c.h"
 
@@ -106,6 +48,9 @@
 
 namespace cv {
 
+//! @addtogroup core
+//! @{
+
 template <typename T> class WImage;
 template <typename T> class WImageBuffer;
 template <typename T> class WImageView;
@@ -165,12 +110,63 @@ typedef WImageC<ushort, 3>        WImage3_16u;
 typedef WImageViewC<ushort, 3>    WImageView3_16u;
 typedef WImageBufferC<ushort, 3>  WImageBuffer3_16u;
 
-//
-// WImage definitions
-//
-// This WImage class gives access to the data it refers to.  It can be
-// constructed either by allocating the data with a WImageBuffer class or
-// using the WImageView class to refer to a subimage or outside data.
+/** @brief Image class which provides a thin layer around an IplImage.
+
+The goals of the class design are:
+
+    -# All the data has explicit ownership to avoid memory leaks
+    -# No hidden allocations or copies for performance.
+    -# Easy access to OpenCV methods (which will access IPP if available)
+    -# Can easily treat external data as an image
+    -# Easy to create images which are subsets of other images
+    -# Fast pixel access which can take advantage of number of channels if known at compile time.
+
+The WImage class is the image class which provides the data accessors. The 'W' comes from the fact
+that it is also a wrapper around the popular but inconvenient IplImage class. A WImage can be
+constructed either using a WImageBuffer class which allocates and frees the data, or using a
+WImageView class which constructs a subimage or a view into external data. The view class does no
+memory management. Each class actually has two versions, one when the number of channels is known
+at compile time and one when it isn't. Using the one with the number of channels specified can
+provide some compile time optimizations by using the fact that the number of channels is a
+constant.
+
+We use the convention (c,r) to refer to column c and row r with (0,0) being the upper left corner.
+This is similar to standard Euclidean coordinates with the first coordinate varying in the
+horizontal direction and the second coordinate varying in the vertical direction. Thus (c,r) is
+usually in the domain [0, width) X [0, height)
+
+Example usage:
+@code
+WImageBuffer3_b  im(5,7);  // Make a 5X7 3 channel image of type uchar
+WImageView3_b  sub_im(im, 2,2, 3,3); // 3X3 submatrix
+vector<float> vec(10, 3.0f);
+WImageView1_f user_im(&vec[0], 2, 5);  // 2X5 image w/ supplied data
+
+im.SetZero();  // same as cvSetZero(im.Ipl())
+*im(2, 3) = 15;  // Modify the element at column 2, row 3
+MySetRand(&sub_im);
+
+// Copy the second row into the first.  This can be done with no memory
+// allocation and will use SSE if IPP is available.
+int w = im.Width();
+im.View(0,0, w,1).CopyFrom(im.View(0,1, w,1));
+
+// Doesn't care about source of data since using WImage
+void MySetRand(WImage_b* im) { // Works with any number of channels
+for (int r = 0; r < im->Height(); ++r) {
+ float* row = im->Row(r);
+ for (int c = 0; c < im->Width(); ++c) {
+    for (int ch = 0; ch < im->Channels(); ++ch, ++row) {
+      *row = uchar(rand() & 255);
+    }
+ }
+}
+}
+@endcode
+
+Functions that are not part of the basic image allocation, viewing, and access should come from
+OpenCV, except some useful functions that are not part of OpenCV can be found in wimage_util.h
+*/
 template<typename T>
 class WImage
 {
@@ -252,10 +248,10 @@ protected:
 };
 
 
-
-// Image class when both the pixel type and number of channels
-// are known at compile time.  This wrapper will speed up some of the operations
-// like accessing individual pixels using the () operator.
+/** Image class when both the pixel type and number of channels
+are known at compile time.  This wrapper will speed up some of the operations
+like accessing individual pixels using the () operator.
+*/
 template<typename T, int C>
 class WImageC : public WImage<T>
 {
@@ -292,12 +288,9 @@ protected:
     }
 };
 
-//
-// WImageBuffer definitions
-//
-// Image class which owns the data, so it can be allocated and is always
-// freed.  It cannot be copied but can be explicity cloned.
-//
+/** Image class which owns the data, so it can be allocated and is always
+freed.  It cannot be copied but can be explicity cloned.
+*/
 template<typename T>
 class WImageBuffer : public WImage<T>
 {
@@ -352,8 +345,8 @@ private:
     void operator=(const WImageBuffer&);
 };
 
-// Like a WImageBuffer class but when the number of channels is known
-// at compile time.
+/** Like a WImageBuffer class but when the number of channels is known at compile time.
+*/
 template<typename T, int C>
 class WImageBufferC : public WImageC<T, C>
 {
@@ -409,14 +402,10 @@ private:
     void operator=(const WImageBufferC&);
 };
 
-//
-// WImageView definitions
-//
-// View into an image class which allows treating a subimage as an image
-// or treating external data as an image
-//
-template<typename T>
-class WImageView : public WImage<T>
+/** View into an image class which allows treating a subimage as an image or treating external data
+as an image
+*/
+template<typename T> class WImageView : public WImage<T>
 {
 public:
     typedef typename WImage<T>::BaseType BaseType;
@@ -518,15 +507,9 @@ inline int WImage<float>::Depth() const {return IPL_DEPTH_32F; }
 template<>
 inline int WImage<double>::Depth() const {return IPL_DEPTH_64F; }
 
-//
-// Pure virtual destructors still need to be defined.
-//
 template<typename T> inline WImage<T>::~WImage() {}
 template<typename T, int C> inline WImageC<T, C>::~WImageC() {}
 
-//
-// Allocate ImageData
-//
 template<typename T>
 inline void WImageBuffer<T>::Allocate(int width, int height, int nchannels)
 {
@@ -547,9 +530,6 @@ inline void WImageBufferC<T, C>::Allocate(int width, int height)
     }
 }
 
-//
-// ImageView methods
-//
 template<typename T>
 WImageView<T>::WImageView(WImage<T>* img, int c, int r, int width, int height)
         : WImage<T>(0)
@@ -614,6 +594,8 @@ WImageViewC<T, C> WImageC<T, C>::View(int c, int r, int width, int height) {
     return WImageViewC<T, C>(this, c, r, width, height);
 }
 
+//! @} core
+
 }  // end of namespace
 
 #endif // __cplusplus