1 /*M/////////////////////////////////////////////////////////////////////////////////////// 2 // 3 // IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING. 4 // 5 // By downloading, copying, installing or using the software you agree to this license. 6 // If you do not agree to this license, do not download, install, 7 // copy or use the software. 8 // 9 // 10 // License Agreement 11 // For Open Source Computer Vision Library 12 // 13 // Copyright (C) 2000-2008, Intel Corporation, all rights reserved. 14 // Copyright (C) 2009, Willow Garage Inc., all rights reserved. 15 // Copyright (C) 2013, OpenCV Foundation, all rights reserved. 16 // Third party copyrights are property of their respective owners. 17 // 18 // Redistribution and use in source and binary forms, with or without modification, 19 // are permitted provided that the following conditions are met: 20 // 21 // * Redistribution's of source code must retain the above copyright notice, 22 // this list of conditions and the following disclaimer. 23 // 24 // * Redistribution's in binary form must reproduce the above copyright notice, 25 // this list of conditions and the following disclaimer in the documentation 26 // and/or other materials provided with the distribution. 27 // 28 // * The name of the copyright holders may not be used to endorse or promote products 29 // derived from this software without specific prior written permission. 30 // 31 // This software is provided by the copyright holders and contributors "as is" and 32 // any express or implied warranties, including, but not limited to, the implied 33 // warranties of merchantability and fitness for a particular purpose are disclaimed. 34 // In no event shall the Intel Corporation or contributors be liable for any direct, 35 // indirect, incidental, special, exemplary, or consequential damages 36 // (including, but not limited to, procurement of substitute goods or services; 37 // loss of use, data, or profits; or business interruption) however caused 38 // and on any theory of liability, whether in contract, strict liability, 39 // or tort (including negligence or otherwise) arising in any way out of 40 // the use of this software, even if advised of the possibility of such damage. 41 // 42 //M*/ 43 44 #pragma once 45 46 #ifndef __OPENCV_CUDEV_GRID_PYR_UP_DETAIL_HPP__ 47 #define __OPENCV_CUDEV_GRID_PYR_UP_DETAIL_HPP__ 48 49 #include "../../common.hpp" 50 #include "../../util/vec_traits.hpp" 51 #include "../../util/saturate_cast.hpp" 52 #include "../../util/type_traits.hpp" 53 #include "../../ptr2d/glob.hpp" 54 #include "../../ptr2d/traits.hpp" 55 56 namespace cv { namespace cudev { 57 58 namespace pyramids_detail 59 { 60 template <class SrcPtr, typename DstType> pyrUp(const SrcPtr src,GlobPtr<DstType> dst,const int src_rows,const int src_cols,const int dst_rows,const int dst_cols)61 __global__ void pyrUp(const SrcPtr src, GlobPtr<DstType> dst, const int src_rows, const int src_cols, const int dst_rows, const int dst_cols) 62 { 63 typedef typename PtrTraits<SrcPtr>::value_type src_type; 64 typedef typename VecTraits<src_type>::elem_type src_elem_type; 65 typedef typename LargerType<float, src_elem_type>::type work_elem_type; 66 typedef typename MakeVec<work_elem_type, VecTraits<src_type>::cn>::type work_type; 67 68 const int x = blockIdx.x * blockDim.x + threadIdx.x; 69 const int y = blockIdx.y * blockDim.y + threadIdx.y; 70 71 __shared__ work_type s_srcPatch[10][10]; 72 __shared__ work_type s_dstPatch[20][16]; 73 74 if (threadIdx.x < 10 && threadIdx.y < 10) 75 { 76 int srcx = static_cast<int>((blockIdx.x * blockDim.x) / 2 + threadIdx.x) - 1; 77 int srcy = static_cast<int>((blockIdx.y * blockDim.y) / 2 + threadIdx.y) - 1; 78 79 srcx = ::abs(srcx); 80 srcx = ::min(src_cols - 1, srcx); 81 82 srcy = ::abs(srcy); 83 srcy = ::min(src_rows - 1, srcy); 84 85 s_srcPatch[threadIdx.y][threadIdx.x] = saturate_cast<work_type>(src(srcy, srcx)); 86 } 87 88 __syncthreads(); 89 90 work_type sum = VecTraits<work_type>::all(0); 91 92 const int evenFlag = static_cast<int>((threadIdx.x & 1) == 0); 93 const int oddFlag = static_cast<int>((threadIdx.x & 1) != 0); 94 const bool eveny = ((threadIdx.y & 1) == 0); 95 const int tidx = threadIdx.x; 96 97 if (eveny) 98 { 99 sum = sum + (evenFlag * 0.0625f) * s_srcPatch[1 + (threadIdx.y >> 1)][1 + ((tidx - 2) >> 1)]; 100 sum = sum + ( oddFlag * 0.25f ) * s_srcPatch[1 + (threadIdx.y >> 1)][1 + ((tidx - 1) >> 1)]; 101 sum = sum + (evenFlag * 0.375f ) * s_srcPatch[1 + (threadIdx.y >> 1)][1 + ((tidx ) >> 1)]; 102 sum = sum + ( oddFlag * 0.25f ) * s_srcPatch[1 + (threadIdx.y >> 1)][1 + ((tidx + 1) >> 1)]; 103 sum = sum + (evenFlag * 0.0625f) * s_srcPatch[1 + (threadIdx.y >> 1)][1 + ((tidx + 2) >> 1)]; 104 } 105 106 s_dstPatch[2 + threadIdx.y][threadIdx.x] = sum; 107 108 if (threadIdx.y < 2) 109 { 110 sum = VecTraits<work_type>::all(0); 111 112 if (eveny) 113 { 114 sum = sum + (evenFlag * 0.0625f) * s_srcPatch[0][1 + ((tidx - 2) >> 1)]; 115 sum = sum + ( oddFlag * 0.25f ) * s_srcPatch[0][1 + ((tidx - 1) >> 1)]; 116 sum = sum + (evenFlag * 0.375f ) * s_srcPatch[0][1 + ((tidx ) >> 1)]; 117 sum = sum + ( oddFlag * 0.25f ) * s_srcPatch[0][1 + ((tidx + 1) >> 1)]; 118 sum = sum + (evenFlag * 0.0625f) * s_srcPatch[0][1 + ((tidx + 2) >> 1)]; 119 } 120 121 s_dstPatch[threadIdx.y][threadIdx.x] = sum; 122 } 123 124 if (threadIdx.y > 13) 125 { 126 sum = VecTraits<work_type>::all(0); 127 128 if (eveny) 129 { 130 sum = sum + (evenFlag * 0.0625f) * s_srcPatch[9][1 + ((tidx - 2) >> 1)]; 131 sum = sum + ( oddFlag * 0.25f ) * s_srcPatch[9][1 + ((tidx - 1) >> 1)]; 132 sum = sum + (evenFlag * 0.375f ) * s_srcPatch[9][1 + ((tidx ) >> 1)]; 133 sum = sum + ( oddFlag * 0.25f ) * s_srcPatch[9][1 + ((tidx + 1) >> 1)]; 134 sum = sum + (evenFlag * 0.0625f) * s_srcPatch[9][1 + ((tidx + 2) >> 1)]; 135 } 136 137 s_dstPatch[4 + threadIdx.y][threadIdx.x] = sum; 138 } 139 140 __syncthreads(); 141 142 sum = VecTraits<work_type>::all(0); 143 144 const int tidy = threadIdx.y; 145 146 sum = sum + 0.0625f * s_dstPatch[2 + tidy - 2][threadIdx.x]; 147 sum = sum + 0.25f * s_dstPatch[2 + tidy - 1][threadIdx.x]; 148 sum = sum + 0.375f * s_dstPatch[2 + tidy ][threadIdx.x]; 149 sum = sum + 0.25f * s_dstPatch[2 + tidy + 1][threadIdx.x]; 150 sum = sum + 0.0625f * s_dstPatch[2 + tidy + 2][threadIdx.x]; 151 152 if (x < dst_cols && y < dst_rows) 153 dst(y, x) = saturate_cast<DstType>(4.0f * sum); 154 } 155 156 template <class SrcPtr, typename DstType> pyrUp(const SrcPtr & src,const GlobPtr<DstType> & dst,int src_rows,int src_cols,int dst_rows,int dst_cols,cudaStream_t stream)157 __host__ void pyrUp(const SrcPtr& src, const GlobPtr<DstType>& dst, int src_rows, int src_cols, int dst_rows, int dst_cols, cudaStream_t stream) 158 { 159 const dim3 block(16, 16); 160 const dim3 grid(divUp(dst_cols, block.x), divUp(dst_rows, block.y)); 161 162 pyrUp<<<grid, block, 0, stream>>>(src, dst, src_rows, src_cols, dst_rows, dst_cols); 163 CV_CUDEV_SAFE_CALL( cudaGetLastError() ); 164 165 if (stream == 0) 166 CV_CUDEV_SAFE_CALL( cudaDeviceSynchronize() ); 167 } 168 } 169 170 }} 171 172 #endif 173