This repository was archived by the owner on Apr 29, 2024. It is now read-only.
Repository navigation
Expand file tree
/
Copy pathoptical_flow_LK.cpp
More file actions
319 lines (268 loc) · 11.9 KB
/
Copy pathoptical_flow_LK.cpp
File metadata and controls
319 lines (268 loc) · 11.9 KB
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
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
// This file is part of VisionCPP, a lightweight C++ template library
// for computer vision and image processing.
//
// Copyright (C) 2016 Codeplay Software Limited. All Rights Reserved.
//
// Contact: [email protected]
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
/// \file optical_flow_LK.cc
///
/// -------Example--------
/// Lucas-Kanade Optical Flow
///
/// \brief This example implements the Lucas-Kanade Optical Flow
/// This optical flow needs to solve the following equation
/// | u | = inv(sum w | Dx^2 Dx*Dy | ) * sum w | -Dx*Dt |
/// | v | ( | Dx*Dy Dy^2 | ) | -Dy*Dt |
/// where w is the 3x3 window of ones
///
/// doing algebraic manipulation
/// n = sum(Dx^2) * sum (Dy^2) - (sum (Dx * Dy)^2)
/// u = (-sum (Dy^2) * sum (Dx*Dt) + sum (Dx*Dy) * sum (Dy*Dt)) / n
/// v = ( sum (Dx*Dt) * sum (Dx*Dy) - sum (Dx^2) * sum (Dy*Dt)) / n
// include OpenCV for camera display
#include <opencv2/opencv.hpp>
// include VisionCPP
#include <visioncpp.hpp>
// operator that transforms uv coordinates into polar coordinates
// it was created for displaying the optical flow in RGB
struct OP_UVtoPolar {
visioncpp::pixel::F32C3 operator()(visioncpp::pixel::F32C2 t) {
float intensity = cl::sycl::clamp(
cl::sycl::sqrt(t[0] * t[0] + t[1] * t[1]) / 2.0f, 0.0f, 1.0f);
float angle = cl::sycl::atan2(t[1], t[0]) / (2.0f * M_PI);
float chn = 1.0f;
return visioncpp::pixel::F32C3(angle, chn, intensity);
}
};
// main program
int main(int argc, char **argv) {
// open video or camera
cv::VideoCapture cap;
if (argc == 1) {
cap.open(0);
std::cout << "To use video" << std::endl;
std::cout << "example>: ./example path/to/video.avi" << std::endl;
} else if (argc > 1) {
cap.open(argv[1]);
}
// check if we succeeded
if (!cap.isOpened()) {
std::cout << "Opening Camera/Video Failed." << std::endl;
return -1;
}
// selecting device using sycl as backend
auto dev = visioncpp::make_device<visioncpp::backend::sycl,
visioncpp::device::cpu>();
// defining the image size constants
constexpr size_t COLS = 640;
constexpr size_t ROWS = 480;
constexpr size_t SM = 16;
// defining sum mask
constexpr size_t N = 15;
constexpr size_t NxN = N * N;
float sum_mask[NxN];
for (size_t i = 0; i < NxN; i++) {
sum_mask[i] = 1.0f;
}
// defining derivatives
float prewitt_x[9] = {-1.0f, 0.0f, 1.0f, -2.0f, 0.0f,
2.0f, -1.0f, 0.0f, 1.0f};
float prewitt_y[9] = {-1.0f, -2.0f, -1.0f, 0.0, 0.0f, 0.0f, 1.0f, 2.0f, 1.0f};
// initializing pointers which will store the final results
std::shared_ptr<uchar> rgbFlow(new uchar[COLS * ROWS * 3],
[](uchar *dataMem) { delete[] dataMem; });
// initializing matrices which will store current and previous frame
cv::Mat current;
cv::Mat previous;
// init matrix which displays in RGB the (u,v) matrix
cv::Mat rgbflow_mat(ROWS, COLS, CV_8UC3, rgbFlow.get());
// flag to control if it is the first frame
bool first = true;
for (;;) {
// don't capture the first frame.
// it needs the previous frame to compute optical flow
if (first) {
// read frame
cap.read(current);
// check if image was loaded
if (!current.data) {
break;
}
// resize image to the desirable size
cv::resize(current, current, cv::Size(COLS, ROWS), 0, 0, cv::INTER_CUBIC);
// set flag to false
first = false;
continue;
}
// Starting building the tree (use {} during the creation of the tree)
{
// copy previous frame
previous = current.clone();
// read frame
cap.read(current);
// check if image was loaded
if (!current.data) {
break;
}
// resize image to the desirable size
cv::resize(current, current, cv::Size(COLS, ROWS), 0, 0, cv::INTER_CUBIC);
// init terminal nodes
auto out =
visioncpp::terminal<visioncpp::pixel::U8C3, COLS, ROWS,
visioncpp::memory_type::Buffer2D>(rgbFlow.get());
auto in =
visioncpp::terminal<visioncpp::pixel::U8C3, COLS, ROWS,
visioncpp::memory_type::Buffer2D>(current.data);
auto prev =
visioncpp::terminal<visioncpp::pixel::U8C3, COLS, ROWS,
visioncpp::memory_type::Buffer2D>(previous.data);
// convert unsigned char to float
auto ifrgb = visioncpp::point_operation<visioncpp::OP_U8C3ToF32C3>(in);
// convert to grey scale previous and current frame
auto ifgrey = visioncpp::point_operation<visioncpp::OP_RGBToGREY>(ifrgb);
// convert unsigned char to float
auto pfrgb = visioncpp::point_operation<visioncpp::OP_U8C3ToF32C3>(prev);
// convert to grey scale previous and current frame
auto pfgrey = visioncpp::point_operation<visioncpp::OP_RGBToGREY>(pfrgb);
// apply derivatives in x, y and t (in current and previous images)
auto px_filter =
visioncpp::terminal<float, 3, 3, visioncpp::memory_type::Buffer2D,
visioncpp::scope::Constant>(prewitt_x);
auto py_filter =
visioncpp::terminal<float, 3, 3, visioncpp::memory_type::Buffer2D,
visioncpp::scope::Constant>(prewitt_y);
// break tree before convolution
auto iifgrey =
visioncpp::schedule<visioncpp::policy::Fuse, SM, SM, SM, SM>(ifgrey);
auto ppfgrey =
visioncpp::schedule<visioncpp::policy::Fuse, SM, SM, SM, SM>(pfgrey);
auto ipx = visioncpp::neighbour_operation<visioncpp::OP_Filter2D>(
iifgrey, px_filter);
auto ipy = visioncpp::neighbour_operation<visioncpp::OP_Filter2D>(
iifgrey, py_filter);
auto ppx = visioncpp::neighbour_operation<visioncpp::OP_Filter2D>(
ppfgrey, px_filter);
auto ppy = visioncpp::neighbour_operation<visioncpp::OP_Filter2D>(
ppfgrey, py_filter);
// break tree after convolution
auto iipx =
visioncpp::schedule<visioncpp::policy::Fuse, SM, SM, SM, SM>(ipx);
auto iipy =
visioncpp::schedule<visioncpp::policy::Fuse, SM, SM, SM, SM>(ipy);
auto pppx =
visioncpp::schedule<visioncpp::policy::Fuse, SM, SM, SM, SM>(ppx);
auto pppy =
visioncpp::schedule<visioncpp::policy::Fuse, SM, SM, SM, SM>(ppy);
// sum derivatives of current and previous frames
auto px = visioncpp::point_operation<visioncpp::OP_Add>(iipx, pppx);
auto py = visioncpp::point_operation<visioncpp::OP_Add>(iipy, pppy);
auto pt = visioncpp::point_operation<visioncpp::OP_Sub>(ifgrey, pfgrey);
auto px2 = visioncpp::point_operation<visioncpp::OP_PowerOf2>(px);
auto py2 = visioncpp::point_operation<visioncpp::OP_PowerOf2>(py);
auto pxy = visioncpp::point_operation<visioncpp::OP_Mul>(px, py);
auto pxt = visioncpp::point_operation<visioncpp::OP_Mul>(px, pt);
auto pyt = visioncpp::point_operation<visioncpp::OP_Mul>(py, pt);
// break tree before convolution
auto ppx2 =
visioncpp::schedule<visioncpp::policy::Fuse, SM, SM, SM, SM>(px2);
auto ppy2 =
visioncpp::schedule<visioncpp::policy::Fuse, SM, SM, SM, SM>(py2);
auto ppxy =
visioncpp::schedule<visioncpp::policy::Fuse, SM, SM, SM, SM>(pxy);
auto ppxt =
visioncpp::schedule<visioncpp::policy::Fuse, SM, SM, SM, SM>(pxt);
auto ppyt =
visioncpp::schedule<visioncpp::policy::Fuse, SM, SM, SM, SM>(pyt);
// Sum neighbours
auto sum_mask_node =
visioncpp::terminal<float, N, N, visioncpp::memory_type::Buffer2D,
visioncpp::scope::Constant>(sum_mask);
auto sumpx2 = visioncpp::neighbour_operation<visioncpp::OP_Filter2D>(
ppx2, sum_mask_node);
auto sumpy2 = visioncpp::neighbour_operation<visioncpp::OP_Filter2D>(
ppy2, sum_mask_node);
auto sumpxy = visioncpp::neighbour_operation<visioncpp::OP_Filter2D>(
ppxy, sum_mask_node);
auto sumpxt = visioncpp::neighbour_operation<visioncpp::OP_Filter2D>(
ppxt, sum_mask_node);
auto sumpyt = visioncpp::neighbour_operation<visioncpp::OP_Filter2D>(
ppyt, sum_mask_node);
// break tree after convolution
auto ksumpx2 =
visioncpp::schedule<visioncpp::policy::Fuse, SM, SM, SM, SM>(sumpx2);
auto ksumpy2 =
visioncpp::schedule<visioncpp::policy::Fuse, SM, SM, SM, SM>(sumpy2);
auto ksumpxy =
visioncpp::schedule<visioncpp::policy::Fuse, SM, SM, SM, SM>(sumpxy);
auto ksumpxt =
visioncpp::schedule<visioncpp::policy::Fuse, SM, SM, SM, SM>(sumpxt);
auto ksumpyt =
visioncpp::schedule<visioncpp::policy::Fuse, SM, SM, SM, SM>(sumpyt);
// now we can finally apply the formulas
// u = (-sum (Dy^2) * sum (Dx*Dt) + sum (Dx*Dy) * sum (Dy*Dt)) / n
// v = ( sum (Dx*Dt) * sum (Dx*Dy) - sum (Dx^2) * sum (Dy*Dt)) / n
// n = sum(Dx^2) * sum (Dy^2) - (sum (Dx * Dy)^2)
auto px2py2 =
visioncpp::point_operation<visioncpp::OP_Mul>(ksumpx2, ksumpy2);
auto pxy2 = visioncpp::point_operation<visioncpp::OP_PowerOf2>(ksumpxy);
auto px2py2_Sub_pxy2 =
visioncpp::point_operation<visioncpp::OP_Sub>(px2py2, pxy2);
auto norm = visioncpp::schedule<visioncpp::policy::Fuse, SM, SM, SM, SM>(
px2py2_Sub_pxy2);
// calculate V
// v = ( sum (Dx*Dt) * sum (Dx*Dy) - sum (Dx^2) * sum (Dy*Dt)) / n
auto pxtpxy =
visioncpp::point_operation<visioncpp::OP_Mul>(ksumpxt, ksumpxy);
auto px2pyt =
visioncpp::point_operation<visioncpp::OP_Mul>(ksumpx2, ksumpyt);
auto pxtpxy_Sub_px2pyt =
visioncpp::point_operation<visioncpp::OP_Sub>(pxtpxy, px2pyt);
auto v = visioncpp::point_operation<visioncpp::OP_Div>(pxtpxy_Sub_px2pyt,
norm);
// calculate U
// u = (-sum (Dy^2) * sum (Dx*Dt) + sum (Dx*Dy) * sum (Dy*Dt)) / n
auto pxypyt =
visioncpp::point_operation<visioncpp::OP_Mul>(ksumpxy, ksumpyt);
auto py2pxt =
visioncpp::point_operation<visioncpp::OP_Mul>(ksumpy2, ksumpxt);
auto pxypft_Sub_py2pxt =
visioncpp::point_operation<visioncpp::OP_Sub>(pxypyt, py2pxt);
auto u = visioncpp::point_operation<visioncpp::OP_Div>(pxypft_Sub_py2pxt,
norm);
// assign result in one matrix with 2 channels
// variable uv contains the optical flow computed for each pixel
auto uv = visioncpp::point_operation<visioncpp::OP_Merge2Chns>(u, v);
// The next operations were created to visualize the optical flow
// convert UV into polar coordinates
auto polar = visioncpp::point_operation<OP_UVtoPolar>(uv);
// convert into RGB
auto frgb = visioncpp::point_operation<visioncpp::OP_HSVToRGB>(polar);
// convert float to char
auto urgb = visioncpp::point_operation<visioncpp::OP_F32C3ToU8C3>(frgb);
// assign the urgb to the output node
auto k = visioncpp::assign(out, urgb);
// execute
visioncpp::execute<visioncpp::policy::Fuse, SM, SM, SM, SM>(k, dev);
}
// display optical flow
cv::imshow("Reference Image", current);
cv::imshow("Optical Flow Lucas-Kanade", rgbflow_mat);
// check button pressed to exit the program
if (cv::waitKey(1) >= 0) break;
}
// release video/camera
cap.release();
return 0;
}