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 // Third party copyrights are property of their respective owners.
16 //
17 // Redistribution and use in source and binary forms, with or without modification,
18 // are permitted provided that the following conditions are met:
19 //
20 //   * Redistribution's of source code must retain the above copyright notice,
21 //     this list of conditions and the following disclaimer.
22 //
23 //   * Redistribution's in binary form must reproduce the above copyright notice,
24 //     this list of conditions and the following disclaimer in the documentation
25 //     and/or other materials provided with the distribution.
26 //
27 //   * The name of the copyright holders may not be used to endorse or promote products
28 //     derived from this software without specific prior written permission.
29 //
30 // This software is provided by the copyright holders and contributors "as is" and
31 // any express or implied warranties, including, but not limited to, the implied
32 // warranties of merchantability and fitness for a particular purpose are disclaimed.
33 // In no event shall the Intel Corporation or contributors be liable for any direct,
34 // indirect, incidental, special, exemplary, or consequential damages
35 // (including, but not limited to, procurement of substitute goods or services;
36 // loss of use, data, or profits; or business interruption) however caused
37 // and on any theory of liability, whether in contract, strict liability,
38 // or tort (including negligence or otherwise) arising in any way out of
39 // the use of this software, even if advised of the possibility of such damage.
40 //
41 //M*/
42 
43 #if !defined CUDA_DISABLER
44 
45 #include "opencv2/core/cuda/common.hpp"
46 #include "opencv2/core/cuda/limits.hpp"
47 #include "opencv2/core/cuda/functional.hpp"
48 #include "opencv2/core/cuda/reduce.hpp"
49 
50 using namespace cv::cuda;
51 using namespace cv::cuda::device;
52 
53 namespace optflowbm
54 {
55     texture<uchar, cudaTextureType2D, cudaReadModeElementType> tex_prev(false, cudaFilterModePoint, cudaAddressModeClamp);
56     texture<uchar, cudaTextureType2D, cudaReadModeElementType> tex_curr(false, cudaFilterModePoint, cudaAddressModeClamp);
57 
cmpBlocks(int X1,int Y1,int X2,int Y2,int2 blockSize)58     __device__ int cmpBlocks(int X1, int Y1, int X2, int Y2, int2 blockSize)
59     {
60         int s = 0;
61 
62         for (int y = 0; y < blockSize.y; ++y)
63         {
64             for (int x = 0; x < blockSize.x; ++x)
65                 s += ::abs(tex2D(tex_prev, X1 + x, Y1 + y) - tex2D(tex_curr, X2 + x, Y2 + y));
66         }
67 
68         return s;
69     }
70 
calcOptFlowBM(PtrStepSzf velx,PtrStepf vely,const int2 blockSize,const int2 shiftSize,const bool usePrevious,const int maxX,const int maxY,const int acceptLevel,const int escapeLevel,const short2 * ss,const int ssCount)71     __global__ void calcOptFlowBM(PtrStepSzf velx, PtrStepf vely, const int2 blockSize, const int2 shiftSize, const bool usePrevious,
72                                   const int maxX, const int maxY, const int acceptLevel, const int escapeLevel,
73                                   const short2* ss, const int ssCount)
74     {
75         const int j = blockIdx.x * blockDim.x + threadIdx.x;
76         const int i = blockIdx.y * blockDim.y + threadIdx.y;
77 
78         if (i >= velx.rows || j >= velx.cols)
79             return;
80 
81         const int X1 = j * shiftSize.x;
82         const int Y1 = i * shiftSize.y;
83 
84         const int offX = usePrevious ? __float2int_rn(velx(i, j)) : 0;
85         const int offY = usePrevious ? __float2int_rn(vely(i, j)) : 0;
86 
87         int X2 = X1 + offX;
88         int Y2 = Y1 + offY;
89 
90         int dist = numeric_limits<int>::max();
91 
92         if (0 <= X2 && X2 <= maxX && 0 <= Y2 && Y2 <= maxY)
93             dist = cmpBlocks(X1, Y1, X2, Y2, blockSize);
94 
95         int countMin = 1;
96         int sumx = offX;
97         int sumy = offY;
98 
99         if (dist > acceptLevel)
100         {
101             // do brute-force search
102             for (int k = 0; k < ssCount; ++k)
103             {
104                 const short2 ssVal = ss[k];
105 
106                 const int dx = offX + ssVal.x;
107                 const int dy = offY + ssVal.y;
108 
109                 X2 = X1 + dx;
110                 Y2 = Y1 + dy;
111 
112                 if (0 <= X2 && X2 <= maxX && 0 <= Y2 && Y2 <= maxY)
113                 {
114                     const int tmpDist = cmpBlocks(X1, Y1, X2, Y2, blockSize);
115                     if (tmpDist < acceptLevel)
116                     {
117                         sumx = dx;
118                         sumy = dy;
119                         countMin = 1;
120                         break;
121                     }
122 
123                     if (tmpDist < dist)
124                     {
125                         dist = tmpDist;
126                         sumx = dx;
127                         sumy = dy;
128                         countMin = 1;
129                     }
130                     else if (tmpDist == dist)
131                     {
132                         sumx += dx;
133                         sumy += dy;
134                         countMin++;
135                     }
136                 }
137             }
138 
139             if (dist > escapeLevel)
140             {
141                 sumx = offX;
142                 sumy = offY;
143                 countMin = 1;
144             }
145         }
146 
147         velx(i, j) = static_cast<float>(sumx) / countMin;
148         vely(i, j) = static_cast<float>(sumy) / countMin;
149     }
150 
calc(PtrStepSzb prev,PtrStepSzb curr,PtrStepSzf velx,PtrStepSzf vely,int2 blockSize,int2 shiftSize,bool usePrevious,int maxX,int maxY,int acceptLevel,int escapeLevel,const short2 * ss,int ssCount,cudaStream_t stream)151     void calc(PtrStepSzb prev, PtrStepSzb curr, PtrStepSzf velx, PtrStepSzf vely, int2 blockSize, int2 shiftSize, bool usePrevious,
152               int maxX, int maxY, int acceptLevel, int escapeLevel, const short2* ss, int ssCount, cudaStream_t stream)
153     {
154         bindTexture(&tex_prev, prev);
155         bindTexture(&tex_curr, curr);
156 
157         const dim3 block(32, 8);
158         const dim3 grid(divUp(velx.cols, block.x), divUp(vely.rows, block.y));
159 
160         calcOptFlowBM<<<grid, block, 0, stream>>>(velx, vely, blockSize, shiftSize, usePrevious,
161                                                   maxX, maxY, acceptLevel,  escapeLevel, ss, ssCount);
162         cudaSafeCall( cudaGetLastError() );
163 
164         if (stream == 0)
165             cudaSafeCall( cudaDeviceSynchronize() );
166     }
167 }
168 
169 #endif // !defined CUDA_DISABLER
170