Skip to content

O
opencv
Commit 17233c68 authored by Alexander Alekhin's avatar Alexander Alekhin
Browse files
Options

Merge pull request #10797 from mshabunin:split-convert

parents 57cbde33 221342fb
Expand all Hide whitespace changes
Inline Side-by-side
Showing with 607 additions and 220 deletions
modules/core/src/channels.cpp 0 → 100644
View file @ 17233c68
This diff is collapsed.
modules/core/src/convert.avx2.cpp
View file @ 17233c68
/*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) 2009-2011, Willow Garage Inc., all rights reserved.
// Copyright (C) 2014-2015, 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*/
// 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
#include "precomp.hpp"
#include "convert.hpp"
......@@ -74,5 +36,5 @@ void cvtScale_s16s32f32Line_AVX2(const short* src, int* dst, float scale, float
}
}
}
} // cv::
/* End of file. */
modules/core/src/convert.cpp
View file @ 17233c68
This diff is collapsed.
modules/core/src/convert.fp16.cpp
View file @ 17233c68
/*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) 2009-2011, Willow Garage Inc., all rights reserved.
// Copyright (C) 2014-2015, 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*/
// 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
#include "precomp.hpp"
#include "precomp.hpp"
#include "convert.hpp"
namespace cv
{
namespace opt_FP16
......@@ -163,5 +123,5 @@ void cvtScaleHalf_SIMD16f32f( const short* src, size_t sstep, float* dst, size_t
#error "Unsupported build configuration"
#endif
}
}
/* End of file. */
} // cv::
modules/core/src/convert.hpp
View file @ 17233c68
// 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
#ifndef SRC_CONVERT_HPP
#define SRC_CONVERT_HPP
#include "opencv2/core/types.hpp"
namespace
{
......@@ -185,4 +194,6 @@ namespace opt_SSE4_1
int cvtScale_SIMD_f64u16f32_SSE41(const double * src, ushort * dst, int width, float scale, float shift);
int Cvt_SIMD_f64u16_SSE41(const double * src, ushort * dst, int width);
}
}
\ No newline at end of file
}
#endif // SRC_CONVERT_HPP
modules/core/src/convert.sse4_1.cpp
View file @ 17233c68
/*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) 2009-2011, Willow Garage Inc., all rights reserved.
// Copyright (C) 2014-2015, 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*/
// 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
#include "precomp.hpp"
#include "convert.hpp"
......@@ -236,6 +198,6 @@ int Cvt_SIMD_f64u16_SSE41(const double * src, ushort * dst, int width)
}
}
}
} // cv::
/* End of file. */
modules/core/src/convert_c.cpp 0 → 100644
View file @ 17233c68
// 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
#include "precomp.hpp"
CV_IMPL void
cvSplit( const void* srcarr, void* dstarr0, void* dstarr1, void* dstarr2, void* dstarr3 )
{
void* dptrs[] = { dstarr0, dstarr1, dstarr2, dstarr3 };
cv::Mat src = cv::cvarrToMat(srcarr);
int i, j, nz = 0;
for( i = 0; i < 4; i++ )
nz += dptrs[i] != 0;
CV_Assert( nz > 0 );
std::vector<cv::Mat> dvec(nz);
std::vector<int> pairs(nz*2);
for( i = j = 0; i < 4; i++ )
{
if( dptrs[i] != 0 )
{
dvec[j] = cv::cvarrToMat(dptrs[i]);
CV_Assert( dvec[j].size() == src.size() );
CV_Assert( dvec[j].depth() == src.depth() );
CV_Assert( dvec[j].channels() == 1 );
CV_Assert( i < src.channels() );
pairs[j*2] = i;
pairs[j*2+1] = j;
j++;
}
}
if( nz == src.channels() )
cv::split( src, dvec );
else
{
cv::mixChannels( &src, 1, &dvec[0], nz, &pairs[0], nz );
}
}
CV_IMPL void
cvMerge( const void* srcarr0, const void* srcarr1, const void* srcarr2,
const void* srcarr3, void* dstarr )
{
const void* sptrs[] = { srcarr0, srcarr1, srcarr2, srcarr3 };
cv::Mat dst = cv::cvarrToMat(dstarr);
int i, j, nz = 0;
for( i = 0; i < 4; i++ )
nz += sptrs[i] != 0;
CV_Assert( nz > 0 );
std::vector<cv::Mat> svec(nz);
std::vector<int> pairs(nz*2);
for( i = j = 0; i < 4; i++ )
{
if( sptrs[i] != 0 )
{
svec[j] = cv::cvarrToMat(sptrs[i]);
CV_Assert( svec[j].size == dst.size &&
svec[j].depth() == dst.depth() &&
svec[j].channels() == 1 && i < dst.channels() );
pairs[j*2] = j;
pairs[j*2+1] = i;
j++;
}
}
if( nz == dst.channels() )
cv::merge( svec, dst );
else
{
cv::mixChannels( &svec[0], nz, &dst, 1, &pairs[0], nz );
}
}
CV_IMPL void
cvMixChannels( const CvArr** src, int src_count,
CvArr** dst, int dst_count,
const int* from_to, int pair_count )
{
cv::AutoBuffer<cv::Mat> buf(src_count + dst_count);
int i;
for( i = 0; i < src_count; i++ )
buf[i] = cv::cvarrToMat(src[i]);
for( i = 0; i < dst_count; i++ )
buf[i+src_count] = cv::cvarrToMat(dst[i]);
cv::mixChannels(&buf[0], src_count, &buf[src_count], dst_count, from_to, pair_count);
}
CV_IMPL void
cvConvertScaleAbs( const void* srcarr, void* dstarr,
double scale, double shift )
{
cv::Mat src = cv::cvarrToMat(srcarr), dst = cv::cvarrToMat(dstarr);
CV_Assert( src.size == dst.size && dst.type() == CV_8UC(src.channels()));
cv::convertScaleAbs( src, dst, scale, shift );
}
CV_IMPL void
cvConvertScale( const void* srcarr, void* dstarr,
double scale, double shift )
{
cv::Mat src = cv::cvarrToMat(srcarr), dst = cv::cvarrToMat(dstarr);
CV_Assert( src.size == dst.size && src.channels() == dst.channels() );
src.convertTo(dst, dst.type(), scale, shift);
}
CV_IMPL void cvLUT( const void* srcarr, void* dstarr, const void* lutarr )
{
cv::Mat src = cv::cvarrToMat(srcarr), dst = cv::cvarrToMat(dstarr), lut = cv::cvarrToMat(lutarr);
CV_Assert( dst.size() == src.size() && dst.type() == CV_MAKETYPE(lut.depth(), src.channels()) );
cv::LUT( src, lut, dst );
}
CV_IMPL void cvNormalize( const CvArr* srcarr, CvArr* dstarr,
double a, double b, int norm_type, const CvArr* maskarr )
{
cv::Mat src = cv::cvarrToMat(srcarr), dst = cv::cvarrToMat(dstarr), mask;
if( maskarr )
mask = cv::cvarrToMat(maskarr);
CV_Assert( dst.size() == src.size() && src.channels() == dst.channels() );
cv::normalize( src, dst, a, b, norm_type, dst.type(), mask );
}
modules/core/src/convert_scale.cpp 0 → 100644
View file @ 17233c68
This diff is collapsed.
modules/core/src/lut.cpp 0 → 100644
View file @ 17233c68
This diff is collapsed.
modules/core/src/merge.cpp
View file @ 17233c68
/*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) 2009-2011, Willow Garage Inc., all rights reserved.
// Copyright (C) 2014-2015, 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*/
// 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
#include "precomp.hpp"
#include "opencl_kernels_core.hpp"
namespace cv { namespace hal {
......@@ -409,4 +372,232 @@ void merge64s(const int64** src, int64* dst, int len, int cn )
merge_(src, dst, len, cn);
}
}}
}} // cv::hal::
typedef void (*MergeFunc)(const uchar** src, uchar* dst, int len, int cn);
static MergeFunc getMergeFunc(int depth)
{
static MergeFunc mergeTab[] =
{
(MergeFunc)GET_OPTIMIZED(cv::hal::merge8u), (MergeFunc)GET_OPTIMIZED(cv::hal::merge8u), (MergeFunc)GET_OPTIMIZED(cv::hal::merge16u), (MergeFunc)GET_OPTIMIZED(cv::hal::merge16u),
(MergeFunc)GET_OPTIMIZED(cv::hal::merge32s), (MergeFunc)GET_OPTIMIZED(cv::hal::merge32s), (MergeFunc)GET_OPTIMIZED(cv::hal::merge64s), 0
};
return mergeTab[depth];
}
#ifdef HAVE_IPP
namespace cv {
static bool ipp_merge(const Mat* mv, Mat& dst, int channels)
{
#ifdef HAVE_IPP_IW
CV_INSTRUMENT_REGION_IPP()
if(channels != 3 && channels != 4)
return false;
if(mv[0].dims <= 2)
{
IppiSize size = ippiSize(mv[0].size());
const void *srcPtrs[4] = {NULL};
size_t srcStep = mv[0].step;
for(int i = 0; i < channels; i++)
{
srcPtrs[i] = mv[i].ptr();
if(srcStep != mv[i].step)
return false;
}
return CV_INSTRUMENT_FUN_IPP(llwiCopyMerge, srcPtrs, (int)srcStep, dst.ptr(), (int)dst.step, size, (int)mv[0].elemSize1(), channels, 0) >= 0;
}
else
{
const Mat *arrays[5] = {NULL};
uchar *ptrs[5] = {NULL};
arrays[0] = &dst;
for(int i = 1; i < channels; i++)
{
arrays[i] = &mv[i-1];
}
NAryMatIterator it(arrays, ptrs);
IppiSize size = { (int)it.size, 1 };
for( size_t i = 0; i < it.nplanes; i++, ++it )
{
if(CV_INSTRUMENT_FUN_IPP(llwiCopyMerge, (const void**)&ptrs[1], 0, ptrs[0], 0, size, (int)mv[0].elemSize1(), channels, 0) < 0)
return false;
}
return true;
}
#else
CV_UNUSED(dst); CV_UNUSED(mv); CV_UNUSED(channels);
return false;
#endif
}
}
#endif
void cv::merge(const Mat* mv, size_t n, OutputArray _dst)
{
CV_INSTRUMENT_REGION()
CV_Assert( mv && n > 0 );
int depth = mv[0].depth();
bool allch1 = true;
int k, cn = 0;
size_t i;
for( i = 0; i < n; i++ )
{
CV_Assert(mv[i].size == mv[0].size && mv[i].depth() == depth);
allch1 = allch1 && mv[i].channels() == 1;
cn += mv[i].channels();
}
CV_Assert( 0 < cn && cn <= CV_CN_MAX );
_dst.create(mv[0].dims, mv[0].size, CV_MAKETYPE(depth, cn));
Mat dst = _dst.getMat();
if( n == 1 )
{
mv[0].copyTo(dst);
return;
}
CV_IPP_RUN_FAST(ipp_merge(mv, dst, (int)n));
if( !allch1 )
{
AutoBuffer<int> pairs(cn*2);
int j, ni=0;
for( i = 0, j = 0; i < n; i++, j += ni )
{
ni = mv[i].channels();
for( k = 0; k < ni; k++ )
{
pairs[(j+k)*2] = j + k;
pairs[(j+k)*2+1] = j + k;
}
}
mixChannels( mv, n, &dst, 1, &pairs[0], cn );
return;
}
MergeFunc func = getMergeFunc(depth);
CV_Assert( func != 0 );
size_t esz = dst.elemSize(), esz1 = dst.elemSize1();
size_t blocksize0 = (int)((BLOCK_SIZE + esz-1)/esz);
AutoBuffer<uchar> _buf((cn+1)*(sizeof(Mat*) + sizeof(uchar*)) + 16);
const Mat** arrays = (const Mat**)(uchar*)_buf;
uchar** ptrs = (uchar**)alignPtr(arrays + cn + 1, 16);
arrays[0] = &dst;
for( k = 0; k < cn; k++ )
arrays[k+1] = &mv[k];
NAryMatIterator it(arrays, ptrs, cn+1);
size_t total = (int)it.size;
size_t blocksize = std::min((size_t)CV_SPLIT_MERGE_MAX_BLOCK_SIZE(cn), cn <= 4 ? total : std::min(total, blocksize0));
for( i = 0; i < it.nplanes; i++, ++it )
{
for( size_t j = 0; j < total; j += blocksize )
{
size_t bsz = std::min(total - j, blocksize);
func( (const uchar**)&ptrs[1], ptrs[0], (int)bsz, cn );
if( j + blocksize < total )
{
ptrs[0] += bsz*esz;
for( int t = 0; t < cn; t++ )
ptrs[t+1] += bsz*esz1;
}
}
}
}
#ifdef HAVE_OPENCL
namespace cv {
static bool ocl_merge( InputArrayOfArrays _mv, OutputArray _dst )
{
std::vector<UMat> src, ksrc;
_mv.getUMatVector(src);
CV_Assert(!src.empty());
int type = src[0].type(), depth = CV_MAT_DEPTH(type),
rowsPerWI = ocl::Device::getDefault().isIntel() ? 4 : 1;
Size size = src[0].size();
for (size_t i = 0, srcsize = src.size(); i < srcsize; ++i)
{
int itype = src[i].type(), icn = CV_MAT_CN(itype), idepth = CV_MAT_DEPTH(itype),
esz1 = CV_ELEM_SIZE1(idepth);
if (src[i].dims > 2)
return false;
CV_Assert(size == src[i].size() && depth == idepth);
for (int cn = 0; cn < icn; ++cn)
{
UMat tsrc = src[i];
tsrc.offset += cn * esz1;
ksrc.push_back(tsrc);
}
}
int dcn = (int)ksrc.size();
String srcargs, processelem, cndecl, indexdecl;
for (int i = 0; i < dcn; ++i)
{
srcargs += format("DECLARE_SRC_PARAM(%d)", i);
processelem += format("PROCESS_ELEM(%d)", i);
indexdecl += format("DECLARE_INDEX(%d)", i);
cndecl += format(" -D scn%d=%d", i, ksrc[i].channels());
}
ocl::Kernel k("merge", ocl::core::split_merge_oclsrc,
format("-D OP_MERGE -D cn=%d -D T=%s -D DECLARE_SRC_PARAMS_N=%s"
" -D DECLARE_INDEX_N=%s -D PROCESS_ELEMS_N=%s%s",
dcn, ocl::memopTypeToStr(depth), srcargs.c_str(),
indexdecl.c_str(), processelem.c_str(), cndecl.c_str()));
if (k.empty())
return false;
_dst.create(size, CV_MAKE_TYPE(depth, dcn));
UMat dst = _dst.getUMat();
int argidx = 0;
for (int i = 0; i < dcn; ++i)
argidx = k.set(argidx, ocl::KernelArg::ReadOnlyNoSize(ksrc[i]));
argidx = k.set(argidx, ocl::KernelArg::WriteOnly(dst));
k.set(argidx, rowsPerWI);
size_t globalsize[2] = { (size_t)dst.cols, ((size_t)dst.rows + rowsPerWI - 1) / rowsPerWI };
return k.run(2, globalsize, NULL, false);
}
}
#endif
void cv::merge(InputArrayOfArrays _mv, OutputArray _dst)
{
CV_INSTRUMENT_REGION()
CV_OCL_RUN(_mv.isUMatVector() && _dst.isUMat(),
ocl_merge(_mv, _dst))
std::vector<Mat> mv;
_mv.getMatVector(mv);
merge(!mv.empty() ? &mv[0] : 0, mv.size(), _dst);
}
modules/core/src/precomp.hpp
View file @ 17233c68
......@@ -126,6 +126,10 @@ template<> inline uchar OpMin<uchar>::operator ()(uchar a, uchar b) const { retu
template<> inline uchar OpMax<uchar>::operator ()(uchar a, uchar b) const { return CV_MAX_8U(a, b); }
typedef void (*UnaryFunc)(const uchar* src1, size_t step1,
uchar* dst, size_t step, Size sz,
void*);
typedef void (*BinaryFunc)(const uchar* src1, size_t step1,
const uchar* src2, size_t step2,
uchar* dst, size_t step, Size sz,
......@@ -137,6 +141,7 @@ typedef void (*BinaryFuncC)(const uchar* src1, size_t step1,
void*);
BinaryFunc getConvertFunc(int sdepth, int ddepth);
BinaryFunc getConvertScaleFunc(int sdepth, int ddepth);
BinaryFunc getCopyMaskFunc(size_t esz);
/* default memory block for sparse array elements */
......@@ -198,6 +203,8 @@ struct NoVec
size_t operator()(const void*, const void*, void*, size_t) const { return 0; }
};
#define CV_SPLIT_MERGE_MAX_BLOCK_SIZE(cn) ((INT_MAX/4)/cn) // HAL implementation accepts 'int' len, so INT_MAX doesn't work here
enum { BLOCK_SIZE = 1024 };
#if defined HAVE_IPP && (IPP_VERSION_X100 >= 700)
......
modules/core/src/split.cpp
View file @ 17233c68
/*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) 2009-2011, Willow Garage Inc., all rights reserved.
// Copyright (C) 2014-2015, 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*/
// 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
#include "precomp.hpp"
#include "opencl_kernels_core.hpp"
namespace cv { namespace hal {
......@@ -425,4 +388,202 @@ void split64s(const int64* src, int64** dst, int len, int cn )
split_(src, dst, len, cn);
}
}}
}} // cv::hal::
/****************************************************************************************\
* split & merge *
\****************************************************************************************/
typedef void (*SplitFunc)(const uchar* src, uchar** dst, int len, int cn);
static SplitFunc getSplitFunc(int depth)
{
static SplitFunc splitTab[] =
{
(SplitFunc)GET_OPTIMIZED(cv::hal::split8u), (SplitFunc)GET_OPTIMIZED(cv::hal::split8u), (SplitFunc)GET_OPTIMIZED(cv::hal::split16u), (SplitFunc)GET_OPTIMIZED(cv::hal::split16u),
(SplitFunc)GET_OPTIMIZED(cv::hal::split32s), (SplitFunc)GET_OPTIMIZED(cv::hal::split32s), (SplitFunc)GET_OPTIMIZED(cv::hal::split64s), 0
};
return splitTab[depth];
}
#ifdef HAVE_IPP
namespace cv {
static bool ipp_split(const Mat& src, Mat* mv, int channels)
{
#ifdef HAVE_IPP_IW
CV_INSTRUMENT_REGION_IPP()
if(channels != 3 && channels != 4)
return false;
if(src.dims <= 2)
{
IppiSize size = ippiSize(src.size());
void *dstPtrs[4] = {NULL};
size_t dstStep = mv[0].step;
for(int i = 0; i < channels; i++)
{
dstPtrs[i] = mv[i].ptr();
if(dstStep != mv[i].step)
return false;
}
return CV_INSTRUMENT_FUN_IPP(llwiCopySplit, src.ptr(), (int)src.step, dstPtrs, (int)dstStep, size, (int)src.elemSize1(), channels, 0) >= 0;
}
else
{
const Mat *arrays[5] = {NULL};
uchar *ptrs[5] = {NULL};
arrays[0] = &src;
for(int i = 1; i < channels; i++)
{
arrays[i] = &mv[i-1];
}
NAryMatIterator it(arrays, ptrs);
IppiSize size = { (int)it.size, 1 };
for( size_t i = 0; i < it.nplanes; i++, ++it )
{
if(CV_INSTRUMENT_FUN_IPP(llwiCopySplit, ptrs[0], 0, (void**)&ptrs[1], 0, size, (int)src.elemSize1(), channels, 0) < 0)
return false;
}
return true;
}
#else
CV_UNUSED(src); CV_UNUSED(mv); CV_UNUSED(channels);
return false;
#endif
}
}
#endif
void cv::split(const Mat& src, Mat* mv)
{
CV_INSTRUMENT_REGION()
int k, depth = src.depth(), cn = src.channels();
if( cn == 1 )
{
src.copyTo(mv[0]);
return;
}
for( k = 0; k < cn; k++ )
{
mv[k].create(src.dims, src.size, depth);
}
CV_IPP_RUN_FAST(ipp_split(src, mv, cn));
SplitFunc func = getSplitFunc(depth);
CV_Assert( func != 0 );
size_t esz = src.elemSize(), esz1 = src.elemSize1();
size_t blocksize0 = (BLOCK_SIZE + esz-1)/esz;
AutoBuffer<uchar> _buf((cn+1)*(sizeof(Mat*) + sizeof(uchar*)) + 16);
const Mat** arrays = (const Mat**)(uchar*)_buf;
uchar** ptrs = (uchar**)alignPtr(arrays + cn + 1, 16);
arrays[0] = &src;
for( k = 0; k < cn; k++ )
{
arrays[k+1] = &mv[k];
}
NAryMatIterator it(arrays, ptrs, cn+1);
size_t total = it.size;
size_t blocksize = std::min((size_t)CV_SPLIT_MERGE_MAX_BLOCK_SIZE(cn), cn <= 4 ? total : std::min(total, blocksize0));
for( size_t i = 0; i < it.nplanes; i++, ++it )
{
for( size_t j = 0; j < total; j += blocksize )
{
size_t bsz = std::min(total - j, blocksize);
func( ptrs[0], &ptrs[1], (int)bsz, cn );
if( j + blocksize < total )
{
ptrs[0] += bsz*esz;
for( k = 0; k < cn; k++ )
ptrs[k+1] += bsz*esz1;
}
}
}
}
#ifdef HAVE_OPENCL
namespace cv {
static bool ocl_split( InputArray _m, OutputArrayOfArrays _mv )
{
int type = _m.type(), depth = CV_MAT_DEPTH(type), cn = CV_MAT_CN(type),
rowsPerWI = ocl::Device::getDefault().isIntel() ? 4 : 1;
String dstargs, processelem, indexdecl;
for (int i = 0; i < cn; ++i)
{
dstargs += format("DECLARE_DST_PARAM(%d)", i);
indexdecl += format("DECLARE_INDEX(%d)", i);
processelem += format("PROCESS_ELEM(%d)", i);
}
ocl::Kernel k("split", ocl::core::split_merge_oclsrc,
format("-D T=%s -D OP_SPLIT -D cn=%d -D DECLARE_DST_PARAMS=%s"
" -D PROCESS_ELEMS_N=%s -D DECLARE_INDEX_N=%s",
ocl::memopTypeToStr(depth), cn, dstargs.c_str(),
processelem.c_str(), indexdecl.c_str()));
if (k.empty())
return false;
Size size = _m.size();
_mv.create(cn, 1, depth);
for (int i = 0; i < cn; ++i)
_mv.create(size, depth, i);
std::vector<UMat> dst;
_mv.getUMatVector(dst);
int argidx = k.set(0, ocl::KernelArg::ReadOnly(_m.getUMat()));
for (int i = 0; i < cn; ++i)
argidx = k.set(argidx, ocl::KernelArg::WriteOnlyNoSize(dst[i]));
k.set(argidx, rowsPerWI);
size_t globalsize[2] = { (size_t)size.width, ((size_t)size.height + rowsPerWI - 1) / rowsPerWI };
return k.run(2, globalsize, NULL, false);
}
}
#endif
void cv::split(InputArray _m, OutputArrayOfArrays _mv)
{
CV_INSTRUMENT_REGION()
CV_OCL_RUN(_m.dims() <= 2 && _mv.isUMatVector(),
ocl_split(_m, _mv))
Mat m = _m.getMat();
if( m.empty() )
{
_mv.release();
return;
}
CV_Assert( !_mv.fixedType() || _mv.empty() || _mv.type() == m.depth() );
int depth = m.depth(), cn = m.channels();
_mv.create(cn, 1, depth);
for (int i = 0; i < cn; ++i)
_mv.create(m.dims, m.size.p, depth, i);
std::vector<Mat> dst;
_mv.getMatVector(dst);
split(m, &dst[0]);
}
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment