1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
//---------------------------------------------------------------------------//
// Copyright (c) 2013-2014 Mageswaran.D <mageswaran1989@gmail.com>
//
// Distributed under the Boost Software License, Version 1.0
// See accompanying file LICENSE_1_0.txt or copy at
// http://www.boost.org/LICENSE_1_0.txt
//
// See http://boostorg.github.com/compute for more information.
//---------------------------------------------------------------------------//
#include <iostream>
#include <string>
#include <opencv2/core/core.hpp>
#include <opencv2/highgui/highgui.hpp>
#include <opencv2/imgproc/imgproc.hpp>
#include <boost/compute/system.hpp>
#include <boost/compute/interop/opencv/core.hpp>
#include <boost/compute/interop/opencv/highgui.hpp>
#include <boost/compute/utility/source.hpp>
#include <boost/program_options.hpp>
namespace compute = boost::compute;
namespace po = boost::program_options;
// Create naive optical flow program
const char source[] = BOOST_COMPUTE_STRINGIZE_SOURCE (
const sampler_t sampler = CLK_ADDRESS_CLAMP_TO_EDGE;
__kernel void optical_flow (
read_only
image2d_t current_image,
read_only image2d_t previous_image,
write_only image2d_t optical_flow,
const float scale,
const float offset,
const float lambda,
const float threshold )
{
int2 coords = (int2)(get_global_id(0), get_global_id(1));
float4 current_pixel = read_imagef(current_image,
sampler,
coords);
float4 previous_pixel = read_imagef(previous_image,
sampler,
coords);
int2 x1 = (int2)(offset, 0.f);
int2 y1 = (int2)(0.f, offset);
//get the difference
float4 curdif = previous_pixel - current_pixel;
//calculate the gradient
//Image 2 first
float4 gradx = read_imagef(previous_image,
sampler,
coords+x1) -
read_imagef(previous_image,
sampler,
coords-x1);
//Image 1
gradx += read_imagef(current_image,
sampler,
coords+x1) -
read_imagef(current_image,
sampler,
coords-x1);
//Image 2 first
float4 grady = read_imagef(previous_image,
sampler,
coords+y1) -
read_imagef(previous_image,
sampler,
coords-y1);
//Image 1
grady += read_imagef(current_image,
sampler,
coords+y1) -
read_imagef(current_image,
sampler,
coords-y1);
float4 sqr = (gradx*gradx) + (grady*grady) +
(float4)(lambda,lambda, lambda, lambda);
float4 gradmag = sqrt(sqr);
///////////////////////////////////////////////////
float4 vx = curdif * (gradx / gradmag);
float vxd = vx.x;//assumes greyscale
//format output for flowrepos, out(-x,+x,-y,+y)
float2 xout = (float2)(fmax(vxd,0.f),fabs(fmin(vxd,0.f)));
xout *= scale;
///////////////////////////////////////////////////
float4 vy = curdif*(grady/gradmag);
float vyd = vy.x;//assumes greyscale
//format output for flowrepos, out(-x,+x,-y,+y)
float2 yout = (float2)(fmax(vyd,0.f),fabs(fmin(vyd,0.f)));
yout *= scale;
///////////////////////////////////////////////////
float4 out = (float4)(xout, yout);
float cond = (float)isgreaterequal(length(out), threshold);
out *= cond;
write_imagef(optical_flow, coords, out);
}
);
// This example shows how to read two images or use camera
// with OpenCV, transfer the frames to the GPU,
// and apply a naive optical flow algorithm
// written in OpenCL
int main(int argc, char *argv[])
{
// setup the command line arguments
po::options_description desc;
desc.add_options()
("help", "show available options")
("camera", po::value<int>()->default_value(-1),
"if not default camera, specify a camera id")
("image1", po::value<std::string>(), "path to image file 1")
("image2", po::value<std::string>(), "path to image file 2");
// Parse the command lines
po::variables_map vm;
po::store(po::parse_command_line(argc, argv, desc), vm);
po::notify(vm);
//check the command line arguments
if(vm.count("help"))
{
std::cout << desc << std::endl;
return 0;
}
//OpenCV variables
cv::Mat previous_cv_image;
cv::Mat current_cv_image;
cv::VideoCapture cap; //OpenCV camera handle
//check for image paths
if(vm.count("image1") && vm.count("image2"))
{
// Read image 1 with OpenCV
previous_cv_image = cv::imread(vm["image1"].as<std::string>(),
CV_LOAD_IMAGE_COLOR);
if(!previous_cv_image.data){
std::cerr << "Failed to load image" << std::endl;
return -1;
}
// Read image 2 with opencv
current_cv_image = cv::imread(vm["image2"].as<std::string>(),
CV_LOAD_IMAGE_COLOR);
if(!current_cv_image.data){
std::cerr << "Failed to load image" << std::endl;
return -1;
}
}
else //by default use camera
{
//open camera
cap.open(vm["camera"].as<int>());
// read first frame
cap >> previous_cv_image;
if(!previous_cv_image.data){
std::cerr << "failed to capture frame" << std::endl;
return -1;
}
// read second frame
cap >> current_cv_image;
if(!current_cv_image.data){
std::cerr << "failed to capture frame" << std::endl;
return -1;
}
}
// Get default device and setup context
compute::device gpu = compute::system::default_device();
compute::context context(gpu);
compute::command_queue queue(context, gpu);
// Convert image to BGRA (OpenCL requires 16-byte aligned data)
cv::cvtColor(previous_cv_image, previous_cv_image, CV_BGR2BGRA);
cv::cvtColor(current_cv_image, current_cv_image, CV_BGR2BGRA);
// Transfer image to gpu
compute::image2d dev_previous_image =
compute::opencv_create_image2d_with_mat(
previous_cv_image, compute::image2d::read_write, queue
);
// Transfer image to gpu
compute::image2d dev_current_image =
compute::opencv_create_image2d_with_mat(
current_cv_image, compute::image2d::read_write, queue
);
// Create output image
compute::image2d dev_output_image(
context,
dev_previous_image.width(),
dev_previous_image.height(),
dev_previous_image.format(),
compute::image2d::write_only
);
compute::program optical_program =
compute::program::create_with_source(source, context);
optical_program.build();
// create flip kernel and set arguments
compute::kernel optical_kernel(optical_program, "optical_flow");
float scale = 10;
float offset = 1;
float lambda = 0.0025;
float threshold = 1.0;
optical_kernel.set_arg(0, dev_previous_image);
optical_kernel.set_arg(1, dev_current_image);
optical_kernel.set_arg(2, dev_output_image);
optical_kernel.set_arg(3, scale);
optical_kernel.set_arg(4, offset);
optical_kernel.set_arg(5, lambda);
optical_kernel.set_arg(6, threshold);
// run flip kernel
size_t origin[2] = { 0, 0 };
size_t region[2] = { dev_previous_image.width(),
dev_previous_image.height() };
queue.enqueue_nd_range_kernel(optical_kernel, 2, origin, region, 0);
//check for image paths
if(vm.count("image1") && vm.count("image2"))
{
// show host image
cv::imshow("Previous Frame", previous_cv_image);
cv::imshow("Current Frame", current_cv_image);
// show gpu image
compute::opencv_imshow("filtered image", dev_output_image, queue);
// wait and return
cv::waitKey(0);
}
else
{
char key = '\0';
while(key != 27) //check for escape key
{
cap >> current_cv_image;
// Convert image to BGRA (OpenCL requires 16-byte aligned data)
cv::cvtColor(current_cv_image, current_cv_image, CV_BGR2BGRA);
// Update the device image memory with current frame data
compute::opencv_copy_mat_to_image(previous_cv_image,
dev_previous_image,
queue);
compute::opencv_copy_mat_to_image(current_cv_image,
dev_current_image,
queue);
// Run the kernel on the device
queue.enqueue_nd_range_kernel(optical_kernel, 2, origin, region, 0);
// Show host image
cv::imshow("Previous Frame", previous_cv_image);
cv::imshow("Current Frame", current_cv_image);
// Show GPU image
compute::opencv_imshow("filtered image", dev_output_image, queue);
// Copy current frame container to previous frame container
current_cv_image.copyTo(previous_cv_image);
// wait
key = cv::waitKey(10);
}
}
return 0;
}