1
/*M///////////////////////////////////////////////////////////////////////////////////////
3
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
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.
11
// For Open Source Computer Vision Library
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.
17
// Redistribution and use in source and binary forms, with or without modification,
18
// are permitted provided that the following conditions are met:
20
// * Redistribution's of source code must retain the above copyright notice,
21
// this list of conditions and the following disclaimer.
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.
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.
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.
43
#if !defined CUDA_DISABLER
45
#include "opencv2/core/cuda/common.hpp"
46
#include "opencv2/core/cuda/limits.hpp"
48
#include "cuda/disparity_bilateral_filter.hpp"
50
namespace cv { namespace cuda { namespace device
52
namespace disp_bilateral_filter
54
template <int channels>
57
static __device__ __forceinline__ uchar calc(const uchar* a, const uchar* b)
59
uchar x = ::abs(a[0] - b[0]);
60
uchar y = ::abs(a[1] - b[1]);
61
uchar z = ::abs(a[2] - b[2]);
62
return (::max(::max(x, y), z));
69
static __device__ __forceinline__ uchar calc(const uchar* a, const uchar* b)
71
return ::abs(a[0] - b[0]);
75
template <int channels, typename T>
76
__global__ void disp_bilateral_filter(int t, T* disp, size_t disp_step,
77
const uchar* img, size_t img_step, int h, int w,
78
const float* ctable_color, const float * ctable_space, size_t ctable_space_step,
80
short cedge_disc, short cmax_disc)
82
const int y = blockIdx.y * blockDim.y + threadIdx.y;
83
const int x = ((blockIdx.x * blockDim.x + threadIdx.x) << 1) + ((y + t) & 1);
87
if (y > 0 && y < h - 1 && x > 0 && x < w - 1)
89
dp[0] = *(disp + (y ) * disp_step + x + 0);
90
dp[1] = *(disp + (y-1) * disp_step + x + 0);
91
dp[2] = *(disp + (y ) * disp_step + x - 1);
92
dp[3] = *(disp + (y+1) * disp_step + x + 0);
93
dp[4] = *(disp + (y ) * disp_step + x + 1);
95
if(::abs(dp[1] - dp[0]) >= cedge_disc || ::abs(dp[2] - dp[0]) >= cedge_disc || ::abs(dp[3] - dp[0]) >= cedge_disc || ::abs(dp[4] - dp[0]) >= cedge_disc)
97
const int ymin = ::max(0, y - cradius);
98
const int xmin = ::max(0, x - cradius);
99
const int ymax = ::min(h - 1, y + cradius);
100
const int xmax = ::min(w - 1, x + cradius);
102
float cost[] = {0.0f, 0.0f, 0.0f, 0.0f, 0.0f};
104
const uchar* ic = img + y * img_step + channels * x;
106
for(int yi = ymin; yi <= ymax; yi++)
108
const T* disp_y = disp + yi * disp_step;
110
for(int xi = xmin; xi <= xmax; xi++)
112
const uchar* in = img + yi * img_step + channels * xi;
114
uchar dist_rgb = DistRgbMax<channels>::calc(in, ic);
116
const float weight = ctable_color[dist_rgb] * (ctable_space + ::abs(y-yi)* ctable_space_step)[::abs(x-xi)];
118
const T disp_reg = disp_y[xi];
120
cost[0] += ::min(cmax_disc, ::abs(disp_reg - dp[0])) * weight;
121
cost[1] += ::min(cmax_disc, ::abs(disp_reg - dp[1])) * weight;
122
cost[2] += ::min(cmax_disc, ::abs(disp_reg - dp[2])) * weight;
123
cost[3] += ::min(cmax_disc, ::abs(disp_reg - dp[3])) * weight;
124
cost[4] += ::min(cmax_disc, ::abs(disp_reg - dp[4])) * weight;
128
float minimum = numeric_limits<float>::max();
131
if (cost[0] < minimum)
136
if (cost[1] < minimum)
141
if (cost[2] < minimum)
146
if (cost[3] < minimum)
151
if (cost[4] < minimum)
157
*(disp + y * disp_step + x) = dp[id];
162
template <typename T>
163
void disp_bilateral_filter(PtrStepSz<T> disp, PtrStepSzb img, int channels, int iters, const float *table_color, const float* table_space, size_t table_step, int radius, short edge_disc, short max_disc, cudaStream_t stream)
165
dim3 threads(32, 8, 1);
167
grid.x = divUp(disp.cols, threads.x << 1);
168
grid.y = divUp(disp.rows, threads.y);
173
for (int i = 0; i < iters; ++i)
175
disp_bilateral_filter<1><<<grid, threads, 0, stream>>>(0, disp.data, disp.step/sizeof(T), img.data, img.step, disp.rows, disp.cols, table_color, table_space, table_step, radius, edge_disc, max_disc);
176
cudaSafeCall( cudaGetLastError() );
178
disp_bilateral_filter<1><<<grid, threads, 0, stream>>>(1, disp.data, disp.step/sizeof(T), img.data, img.step, disp.rows, disp.cols, table_color, table_space, table_step, radius, edge_disc, max_disc);
179
cudaSafeCall( cudaGetLastError() );
183
for (int i = 0; i < iters; ++i)
185
disp_bilateral_filter<3><<<grid, threads, 0, stream>>>(0, disp.data, disp.step/sizeof(T), img.data, img.step, disp.rows, disp.cols, table_color, table_space, table_step, radius, edge_disc, max_disc);
186
cudaSafeCall( cudaGetLastError() );
188
disp_bilateral_filter<3><<<grid, threads, 0, stream>>>(1, disp.data, disp.step/sizeof(T), img.data, img.step, disp.rows, disp.cols, table_color, table_space, table_step, radius, edge_disc, max_disc);
189
cudaSafeCall( cudaGetLastError() );
193
CV_Error(cv::Error::BadNumChannels, "Unsupported channels count");
197
cudaSafeCall( cudaDeviceSynchronize() );
200
template void disp_bilateral_filter<uchar>(PtrStepSz<uchar> disp, PtrStepSzb img, int channels, int iters, const float *table_color, const float *table_space, size_t table_step, int radius, short, short, cudaStream_t stream);
201
template void disp_bilateral_filter<short>(PtrStepSz<short> disp, PtrStepSzb img, int channels, int iters, const float *table_color, const float *table_space, size_t table_step, int radius, short, short, cudaStream_t stream);
202
} // namespace bilateral_filter
203
}}} // namespace cv { namespace cuda { namespace cudev
205
#endif /* CUDA_DISABLER */