Skip to content

O
opencv
Commit e1e5047b authored by Vladislav Vinogradov's avatar Vladislav Vinogradov
Browse files
Options

added gpu::LUT for CV_8UC3 type, added gpu::cvtColor for BGR2BGR5x5, minor fix in tests.

parent 1b8c0000
Expand all Hide whitespace changes
Inline Side-by-side
Showing with 82 additions and 47 deletions
modules/gpu/src/arithm.cpp
View file @ e1e5047b
......@@ -266,13 +266,13 @@ double cv::gpu::norm(const GpuMat& src1, const GpuMat& src2, int normType)
sz.height = src1.rows;
int funcIdx = normType >> 1;
Scalar retVal;
double retVal;
nppSafeCall( npp_norm_diff_func[funcIdx](src1.ptr<Npp8u>(), src1.step,
src2.ptr<Npp8u>(), src2.step,
sz, retVal.val) );
sz, &retVal) );
return retVal[0];
return retVal;
}
////////////////////////////////////////////////////////////////////////
......@@ -307,10 +307,7 @@ void cv::gpu::flip(const GpuMat& src, GpuMat& dst, int flipCode)
Scalar cv::gpu::sum(const GpuMat& src)
{
CV_Assert(src.type() == CV_8UC1 || src.type() == CV_8UC4);
CV_Assert(src.type() == CV_8UC1 || src.type() == CV_8UC4);
NppiSize sz;
sz.width = src.cols;
......@@ -324,7 +321,7 @@ Scalar cv::gpu::sum(const GpuMat& src)
GpuMat buf(1, bufsz, CV_32S);
Scalar res;
nppSafeCall( nppiSum_8u_C1R(src.ptr<Npp8u>(), src.step, sz, buf.ptr<Npp32s>(), res.val) );
nppSafeCall( nppiSum_8u_C1R(src.ptr<Npp8u>(), src.step, sz, buf.ptr<Npp32s>(), res.val) );
return res;
}
else
......@@ -336,8 +333,6 @@ Scalar cv::gpu::sum(const GpuMat& src)
nppSafeCall( nppiSum_8u_C4R(src.ptr<Npp8u>(), src.step, sz, buf.ptr<Npp32s>(), res.val) );
return res;
}
}
////////////////////////////////////////////////////////////////////////
......@@ -371,28 +366,54 @@ void cv::gpu::LUT(const GpuMat& src, const Mat& lut, GpuMat& dst)
{
public:
Npp32s pLevels[256];
const Npp32s* pLevels3[3];
int nValues3[3];
LevelsInit()
{
{
nValues3[0] = nValues3[1] = nValues3[2] = 256;
for (int i = 0; i < 256; ++i)
pLevels[i] = i;
pLevels3[0] = pLevels3[1] = pLevels3[2] = pLevels;
}
};
static LevelsInit lvls;
int cn = src.channels();
CV_Assert(src.type() == CV_8UC1);
CV_Assert(lut.depth() == CV_32SC1 && lut.rows * lut.cols == 256 && lut.isContinuous());
CV_Assert(src.type() == CV_8UC1 || src.type() == CV_8UC3);
CV_Assert(lut.depth() == CV_8U && (lut.channels() == 1 || lut.channels() == cn) && lut.rows * lut.cols == 256 && lut.isContinuous());
dst.create(src.size(), src.type());
dst.create(src.size(), CV_MAKETYPE(lut.depth(), cn));
NppiSize sz;
sz.height = src.rows;
sz.width = src.cols;
Mat nppLut;
lut.convertTo(nppLut, CV_32S);
nppSafeCall( nppiLUT_Linear_8u_C1R(src.ptr<Npp8u>(), src.step, dst.ptr<Npp8u>(), dst.step, sz,
lut.ptr<Npp32s>(), lvls.pLevels, 256) );
if (src.type() == CV_8UC1)
{
nppSafeCall( nppiLUT_Linear_8u_C1R(src.ptr<Npp8u>(), src.step, dst.ptr<Npp8u>(), dst.step, sz,
nppLut.ptr<Npp32s>(), lvls.pLevels, 256) );
}
else
{
Mat nppLut3[3];
const Npp32s* pValues3[3];
if (nppLut.channels() == 1)
pValues3[0] = pValues3[1] = pValues3[2] = nppLut.ptr<Npp32s>();
else
{
cv::split(nppLut, nppLut3);
pValues3[0] = nppLut3[0].ptr<Npp32s>();
pValues3[1] = nppLut3[1].ptr<Npp32s>();
pValues3[2] = nppLut3[2].ptr<Npp32s>();
}
nppSafeCall( nppiLUT_Linear_8u_C3R(src.ptr<Npp8u>(), src.step, dst.ptr<Npp8u>(), dst.step, sz,
pValues3, lvls.pLevels3, lvls.nValues3) );
}
}
#endif /* !defined (HAVE_CUDA) */
\ No newline at end of file
modules/gpu/src/cuda/color.cu
View file @ e1e5047b
This diff is collapsed.
modules/gpu/src/imgproc_gpu.cpp
View file @ e1e5047b
......@@ -81,13 +81,16 @@ namespace cv { namespace gpu
void reprojectImageTo3D_gpu(const DevMem2D& disp, const DevMem2Df& xyzw, const float* q, const cudaStream_t& stream);
void reprojectImageTo3D_gpu(const DevMem2D_<short>& disp, const DevMem2Df& xyzw, const float* q, const cudaStream_t& stream);
void swapChannels_gpu(const DevMem2D& src, const DevMem2D& dst, int cn, const int* coeffs, cudaStream_t stream);
void swapChannels_gpu(const DevMem2D_<ushort>& src, const DevMem2D_<ushort>& dst, int cn, const int* coeffs, cudaStream_t stream);
void swapChannels_gpu(const DevMem2Df& src, const DevMem2Df& dst, int cn, const int* coeffs, cudaStream_t stream);
void swapChannels_gpu_8u(const DevMem2D& src, const DevMem2D& dst, int cn, const int* coeffs, cudaStream_t stream);
void swapChannels_gpu_16u(const DevMem2D& src, const DevMem2D& dst, int cn, const int* coeffs, cudaStream_t stream);
void swapChannels_gpu_32f(const DevMem2D& src, const DevMem2D& dst, int cn, const int* coeffs, cudaStream_t stream);
void RGB2RGB_gpu(const DevMem2D& src, int srccn, const DevMem2D& dst, int dstcn, int bidx, cudaStream_t stream);
void RGB2RGB_gpu(const DevMem2D_<ushort>& src, int srccn, const DevMem2D_<ushort>& dst, int dstcn, int bidx, cudaStream_t stream);
void RGB2RGB_gpu(const DevMem2Df& src, int srccn, const DevMem2Df& dst, int dstcn, int bidx, cudaStream_t stream);
void RGB2RGB_gpu_8u(const DevMem2D& src, int srccn, const DevMem2D& dst, int dstcn, int bidx, cudaStream_t stream);
void RGB2RGB_gpu_16u(const DevMem2D& src, int srccn, const DevMem2D& dst, int dstcn, int bidx, cudaStream_t stream);
void RGB2RGB_gpu_32f(const DevMem2D& src, int srccn, const DevMem2D& dst, int dstcn, int bidx, cudaStream_t stream);
void RGB5x52RGB_gpu(const DevMem2D& src, int green_bits, const DevMem2D& dst, int dstcn, int bidx, cudaStream_t stream);
void RGB2RGB5x5_gpu(const DevMem2D& src, int srccn, const DevMem2D& dst, int green_bits, int bidx, cudaStream_t stream);
void Gray2RGB_gpu(const DevMem2D& src, const DevMem2D& dst, int dstcn, cudaStream_t stream);
void Gray2RGB_gpu(const DevMem2D_<ushort>& src, const DevMem2D_<ushort>& dst, int dstcn, cudaStream_t stream);
......@@ -245,38 +248,36 @@ namespace
out.create(sz, CV_MAKETYPE(depth, dcn));
if( depth == CV_8U )
improc::RGB2RGB_gpu((DevMem2D)src, scn, (DevMem2D)out, dcn, bidx, stream);
improc::RGB2RGB_gpu_8u(src, scn, out, dcn, bidx, stream);
else if( depth == CV_16U )
improc::RGB2RGB_gpu((DevMem2D_<unsigned short>)src, scn, (DevMem2D_<unsigned short>)out, dcn, bidx, stream);
improc::RGB2RGB_gpu_16u(src, scn, out, dcn, bidx, stream);
else
improc::RGB2RGB_gpu((DevMem2Df)src, scn, (DevMem2Df)out, dcn, bidx, stream);
improc::RGB2RGB_gpu_32f(src, scn, out, dcn, bidx, stream);
break;
//case CV_BGR2BGR565: case CV_BGR2BGR555: case CV_RGB2BGR565: case CV_RGB2BGR555:
//case CV_BGRA2BGR565: case CV_BGRA2BGR555: case CV_RGBA2BGR565: case CV_RGBA2BGR555:
// CV_Assert( (scn == 3 || scn == 4) && depth == CV_8U );
// dst.create(sz, CV_8UC2);
//
// CvtColorLoop(src, dst, RGB2RGB5x5(scn,
// code == CV_BGR2BGR565 || code == CV_BGR2BGR555 ||
// code == CV_BGRA2BGR565 || code == CV_BGRA2BGR555 ? 0 : 2,
// code == CV_BGR2BGR565 || code == CV_RGB2BGR565 ||
// code == CV_BGRA2BGR565 || code == CV_RGBA2BGR565 ? 6 : 5 // green bits
// ));
// break;
case CV_BGR2BGR565: case CV_BGR2BGR555: case CV_RGB2BGR565: case CV_RGB2BGR555:
case CV_BGRA2BGR565: case CV_BGRA2BGR555: case CV_RGBA2BGR565: case CV_RGBA2BGR555:
CV_Assert( (scn == 3 || scn == 4) && depth == CV_8U );
out.create(sz, CV_8UC2);
improc::RGB2RGB5x5_gpu(src, scn, out, code == CV_BGR2BGR565 || code == CV_RGB2BGR565 ||
code == CV_BGRA2BGR565 || code == CV_RGBA2BGR565 ? 6 : 5,
code == CV_BGR2BGR565 || code == CV_BGR2BGR555 ||
code == CV_BGRA2BGR565 || code == CV_BGRA2BGR555 ? 0 : 2,
stream);
break;
//case CV_BGR5652BGR: case CV_BGR5552BGR: case CV_BGR5652RGB: case CV_BGR5552RGB:
//case CV_BGR5652BGRA: case CV_BGR5552BGRA: case CV_BGR5652RGBA: case CV_BGR5552RGBA:
// if(dcn <= 0) dcn = 3;
// CV_Assert( (dcn == 3 || dcn == 4) && scn == 2 && depth == CV_8U );
// dst.create(sz, CV_MAKETYPE(depth, dcn));
//
// CvtColorLoop(src, dst, RGB5x52RGB(dcn,
// code == CV_BGR5652BGR || code == CV_BGR5552BGR ||
// code == CV_BGR5652BGRA || code == CV_BGR5552BGRA ? 0 : 2, // blue idx
// code == CV_BGR5652BGR || code == CV_BGR5652RGB ||
// code == CV_BGR5652BGRA || code == CV_BGR5652RGBA ? 6 : 5 // green bits
// ));
// out.create(sz, CV_MAKETYPE(depth, dcn));
// improc::RGB5x52RGB_gpu(src, code == CV_BGR2BGR565 || code == CV_RGB2BGR565 ||
// code == CV_BGRA2BGR565 || code == CV_RGBA2BGR565 ? 6 : 5, out, dcn,
// code == CV_BGR2BGR565 || code == CV_BGR2BGR555 ||
// code == CV_BGRA2BGR565 || code == CV_BGRA2BGR555 ? 0 : 2,
// stream);
// break;
case CV_BGR2GRAY: case CV_BGRA2GRAY: case CV_RGB2GRAY: case CV_RGBA2GRAY:
......@@ -329,7 +330,7 @@ namespace
nppSafeCall( nppiRGBToYCbCr_8u_C3R(src.ptr<Npp8u>(), src.step, out.ptr<Npp8u>(), out.step, nppsz) );
{
static int coeffs[] = {0, 2, 1};
improc::swapChannels_gpu((DevMem2D)out, (DevMem2D)out, 3, coeffs, 0);
improc::swapChannels_gpu_8u(out, out, 3, coeffs, 0);
}
break;
......@@ -341,7 +342,7 @@ namespace
{
static int coeffs[] = {0, 2, 1};
GpuMat src1(src.size(), src.type());
improc::swapChannels_gpu((DevMem2D)src, (DevMem2D)src1, 3, coeffs, 0);
improc::swapChannels_gpu_8u(src, src1, 3, coeffs, 0);
nppSafeCall( nppiYCbCrToRGB_8u_C3R(src1.ptr<Npp8u>(), src1.step, out.ptr<Npp8u>(), out.step, nppsz) );
}
break;
......
tests/gpu/src/arithm.cpp
View file @ e1e5047b
This diff is collapsed.
tests/gpu/src/gputest_main.cpp
View file @ e1e5047b
......@@ -46,6 +46,18 @@ CvTS test_system;
const char* blacklist[] =
{
"GPU-NppImageSum",
"GPU-MatOperatorAsyncCall",
//"GPU-NppErode",
//"GPU-NppDilate",
//"GPU-NppMorphologyEx",
//"GPU-NppImageDivide",
//"GPU-NppImageMeanStdDev",
//"GPU-NppImageMinNax",
//"GPU-NppImageResize",
//"GPU-NppImageWarpAffine",
//"GPU-NppImageWarpPerspective",
//"GPU-NppImageIntegral",
//"GPU-NppImageBlur",
0
};
......
tests/gpu/src/imgproc_gpu.cpp
View file @ e1e5047b
This diff is collapsed.
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