funcs[src.depth()][src.channels() - 1](src, dst, flipCode, StreamAccessor::getStream(stream));
}
-////////////////////////////////////////////////////////////////////////
-// LUT
-
-#if (CUDA_VERSION >= 5000)
-
-namespace
-{
- class LookUpTableImpl : public LookUpTable
- {
- public:
- LookUpTableImpl(InputArray lut);
-
- void transform(InputArray src, OutputArray dst, Stream& stream = Stream::Null());
-
- private:
- int lut_cn;
-
- int nValues3[3];
- const Npp32s* pValues3[3];
- const Npp32s* pLevels3[3];
-
- GpuMat d_pLevels;
- GpuMat d_nppLut;
- GpuMat d_nppLut3[3];
- };
-
- LookUpTableImpl::LookUpTableImpl(InputArray _lut)
- {
- nValues3[0] = nValues3[1] = nValues3[2] = 256;
-
- Npp32s pLevels[256];
- for (int i = 0; i < 256; ++i)
- pLevels[i] = i;
-
- d_pLevels.upload(Mat(1, 256, CV_32S, pLevels));
- pLevels3[0] = pLevels3[1] = pLevels3[2] = d_pLevels.ptr<Npp32s>();
-
- GpuMat lut;
- if (_lut.kind() == _InputArray::GPU_MAT)
- {
- lut = _lut.getGpuMat();
- }
- else
- {
- Mat hLut = _lut.getMat();
- CV_Assert( hLut.total() == 256 && hLut.isContinuous() );
- lut.upload(Mat(1, 256, hLut.type(), hLut.data));
- }
-
- lut_cn = lut.channels();
-
- CV_Assert( lut.depth() == CV_8U );
- CV_Assert( lut.rows == 1 && lut.cols == 256 );
-
- lut.convertTo(d_nppLut, CV_32S);
-
- if (lut_cn == 1)
- {
- pValues3[0] = pValues3[1] = pValues3[2] = d_nppLut.ptr<Npp32s>();
- }
- else
- {
- cuda::split(d_nppLut, d_nppLut3);
-
- pValues3[0] = d_nppLut3[0].ptr<Npp32s>();
- pValues3[1] = d_nppLut3[1].ptr<Npp32s>();
- pValues3[2] = d_nppLut3[2].ptr<Npp32s>();
- }
- }
-
- void LookUpTableImpl::transform(InputArray _src, OutputArray _dst, Stream& _stream)
- {
- GpuMat src = _src.getGpuMat();
-
- const int cn = src.channels();
-
- CV_Assert( src.type() == CV_8UC1 || src.type() == CV_8UC3 );
- CV_Assert( lut_cn == 1 || lut_cn == cn );
-
- _dst.create(src.size(), src.type());
- GpuMat dst = _dst.getGpuMat();
-
- cudaStream_t stream = StreamAccessor::getStream(_stream);
-
- NppStreamHandler h(stream);
-
- NppiSize sz;
- sz.height = src.rows;
- sz.width = src.cols;
-
- if (src.type() == CV_8UC1)
- {
- nppSafeCall( nppiLUT_Linear_8u_C1R(src.ptr<Npp8u>(), static_cast<int>(src.step),
- dst.ptr<Npp8u>(), static_cast<int>(dst.step), sz, d_nppLut.ptr<Npp32s>(), d_pLevels.ptr<Npp32s>(), 256) );
- }
- else
- {
- nppSafeCall( nppiLUT_Linear_8u_C3R(src.ptr<Npp8u>(), static_cast<int>(src.step),
- dst.ptr<Npp8u>(), static_cast<int>(dst.step), sz, pValues3, pLevels3, nValues3) );
- }
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
- }
-}
-
-#else // (CUDA_VERSION >= 5000)
-
-namespace
-{
- class LookUpTableImpl : public LookUpTable
- {
- public:
- LookUpTableImpl(InputArray lut);
-
- void transform(InputArray src, OutputArray dst, Stream& stream = Stream::Null());
-
- private:
- int lut_cn;
-
- Npp32s pLevels[256];
- int nValues3[3];
- const Npp32s* pValues3[3];
- const Npp32s* pLevels3[3];
-
- Mat nppLut;
- Mat nppLut3[3];
- };
-
- LookUpTableImpl::LookUpTableImpl(InputArray _lut)
- {
- nValues3[0] = nValues3[1] = nValues3[2] = 256;
-
- for (int i = 0; i < 256; ++i)
- pLevels[i] = i;
- pLevels3[0] = pLevels3[1] = pLevels3[2] = pLevels;
-
- Mat lut;
- if (_lut.kind() == _InputArray::GPU_MAT)
- {
- lut = Mat(_lut.getGpuMat());
- }
- else
- {
- Mat hLut = _lut.getMat();
- CV_Assert( hLut.total() == 256 && hLut.isContinuous() );
- lut = hLut;
- }
-
- lut_cn = lut.channels();
-
- CV_Assert( lut.depth() == CV_8U );
- CV_Assert( lut.rows == 1 && lut.cols == 256 );
-
- lut.convertTo(nppLut, CV_32S);
-
- if (lut_cn == 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>();
- }
- }
-
- void LookUpTableImpl::transform(InputArray _src, OutputArray _dst, Stream& _stream)
- {
- GpuMat src = _src.getGpuMat();
-
- const int cn = src.channels();
-
- CV_Assert( src.type() == CV_8UC1 || src.type() == CV_8UC3 );
- CV_Assert( lut_cn == 1 || lut_cn == cn );
-
- _dst.create(src.size(), src.type());
- GpuMat dst = _dst.getGpuMat();
-
- cudaStream_t stream = StreamAccessor::getStream(_stream);
-
- NppStreamHandler h(stream);
-
- NppiSize sz;
- sz.height = src.rows;
- sz.width = src.cols;
-
- if (src.type() == CV_8UC1)
- {
- nppSafeCall( nppiLUT_Linear_8u_C1R(src.ptr<Npp8u>(), static_cast<int>(src.step),
- dst.ptr<Npp8u>(), static_cast<int>(dst.step), sz, nppLut.ptr<Npp32s>(), pLevels, 256) );
- }
- else
- {
- nppSafeCall( nppiLUT_Linear_8u_C3R(src.ptr<Npp8u>(), static_cast<int>(src.step),
- dst.ptr<Npp8u>(), static_cast<int>(dst.step), sz, pValues3, pLevels3, nValues3) );
- }
-
- if (stream == 0)
- cudaSafeCall( cudaDeviceSynchronize() );
- }
-}
-
-#endif // (CUDA_VERSION >= 5000)
-
-Ptr<LookUpTable> cv::cuda::createLookUpTable(InputArray lut)
-{
- return makePtr<LookUpTableImpl>(lut);
-}
-
#endif /* !defined (HAVE_CUDA) */
--- /dev/null
+/*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*/
+
+#include "opencv2/opencv_modules.hpp"
+
+#ifndef HAVE_OPENCV_CUDEV
+
+#error "opencv_cudev is required"
+
+#else
+
+#include "opencv2/cudaarithm.hpp"
+#include "opencv2/cudev.hpp"
+
+using namespace cv;
+using namespace cv::cudev;
+
+namespace
+{
+ texture<uchar, cudaTextureType1D, cudaReadModeElementType> texLutTable;
+
+ class LookUpTableImpl : public LookUpTable
+ {
+ public:
+ LookUpTableImpl(InputArray lut);
+ ~LookUpTableImpl();
+
+ void transform(InputArray src, OutputArray dst, Stream& stream = Stream::Null());
+
+ private:
+ GpuMat d_lut;
+ cudaTextureObject_t texLutTableObj;
+ bool cc30;
+ };
+
+ LookUpTableImpl::LookUpTableImpl(InputArray _lut)
+ {
+ if (_lut.kind() == _InputArray::GPU_MAT)
+ {
+ d_lut = _lut.getGpuMat();
+ }
+ else
+ {
+ Mat h_lut = _lut.getMat();
+ d_lut.upload(Mat(1, 256, h_lut.type(), h_lut.data));
+ }
+
+ CV_Assert( d_lut.depth() == CV_8U );
+ CV_Assert( d_lut.rows == 1 && d_lut.cols == 256 );
+
+ cc30 = deviceSupports(FEATURE_SET_COMPUTE_30);
+
+ if (cc30)
+ {
+ // Use the texture object
+ cudaResourceDesc texRes;
+ std::memset(&texRes, 0, sizeof(texRes));
+ texRes.resType = cudaResourceTypeLinear;
+ texRes.res.linear.devPtr = d_lut.data;
+ texRes.res.linear.desc = cudaCreateChannelDesc<uchar>();
+ texRes.res.linear.sizeInBytes = 256 * d_lut.channels() * sizeof(uchar);
+
+ cudaTextureDesc texDescr;
+ std::memset(&texDescr, 0, sizeof(texDescr));
+
+ CV_CUDEV_SAFE_CALL( cudaCreateTextureObject(&texLutTableObj, &texRes, &texDescr, 0) );
+ }
+ else
+ {
+ // Use the texture reference
+ cudaChannelFormatDesc desc = cudaCreateChannelDesc<uchar>();
+ CV_CUDEV_SAFE_CALL( cudaBindTexture(0, &texLutTable, d_lut.data, &desc) );
+ }
+ }
+
+ LookUpTableImpl::~LookUpTableImpl()
+ {
+ if (cc30)
+ {
+ // Use the texture object
+ cudaDestroyTextureObject(texLutTableObj);
+ }
+ else
+ {
+ // Use the texture reference
+ cudaUnbindTexture(texLutTable);
+ }
+ }
+
+ struct LutTablePtrC1
+ {
+ typedef uchar value_type;
+ typedef uchar index_type;
+
+ cudaTextureObject_t texLutTableObj;
+
+ __device__ __forceinline__ uchar operator ()(uchar, uchar x) const
+ {
+ #if CV_CUDEV_ARCH < 300
+ // Use the texture reference
+ return tex1Dfetch(texLutTable, x);
+ #else
+ // Use the texture object
+ return tex1Dfetch<uchar>(texLutTableObj, x);
+ #endif
+ }
+ };
+ struct LutTablePtrC3
+ {
+ typedef uchar3 value_type;
+ typedef uchar3 index_type;
+
+ cudaTextureObject_t texLutTableObj;
+
+ __device__ __forceinline__ uchar3 operator ()(const uchar3&, const uchar3& x) const
+ {
+ #if CV_CUDEV_ARCH < 300
+ // Use the texture reference
+ return make_uchar3(tex1Dfetch(texLutTable, x.x * 3), tex1Dfetch(texLutTable, x.y * 3 + 1), tex1Dfetch(texLutTable, x.z * 3 + 2));
+ #else
+ // Use the texture object
+ return make_uchar3(tex1Dfetch<uchar>(texLutTableObj, x.x * 3), tex1Dfetch<uchar>(texLutTableObj, x.y * 3 + 1), tex1Dfetch<uchar>(texLutTableObj, x.z * 3 + 2));
+ #endif
+ }
+ };
+
+ void LookUpTableImpl::transform(InputArray _src, OutputArray _dst, Stream& stream)
+ {
+ GpuMat src = _src.getGpuMat();
+
+ const int cn = src.channels();
+ const int lut_cn = d_lut.channels();
+
+ CV_Assert( src.type() == CV_8UC1 || src.type() == CV_8UC3 );
+ CV_Assert( lut_cn == 1 || lut_cn == cn );
+
+ _dst.create(src.size(), src.type());
+ GpuMat dst = _dst.getGpuMat();
+
+ if (lut_cn == 1)
+ {
+ GpuMat_<uchar> src1(src.reshape(1));
+ GpuMat_<uchar> dst1(dst.reshape(1));
+
+ LutTablePtrC1 tbl;
+ tbl.texLutTableObj = texLutTableObj;
+
+ dst1.assign(lut_(src1, tbl), stream);
+ }
+ else if (lut_cn == 3)
+ {
+ GpuMat_<uchar3>& src3 = (GpuMat_<uchar3>&) src;
+ GpuMat_<uchar3>& dst3 = (GpuMat_<uchar3>&) dst;
+
+ LutTablePtrC3 tbl;
+ tbl.texLutTableObj = texLutTableObj;
+
+ dst3.assign(lut_(src3, tbl), stream);
+ }
+ }
+}
+
+Ptr<LookUpTable> cv::cuda::createLookUpTable(InputArray lut)
+{
+ return makePtr<LookUpTableImpl>(lut);
+}
+
+#endif
projects / profile / ivi / opencv.git / commitdiff