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 | #include "opencv2/core/types.hpp" |
44 | #include "precomp.hpp" |
45 | #include "distortion_model.hpp" |
46 | |
47 | #include "calib3d_c_api.h" |
48 | |
49 | #include "undistort.simd.hpp" |
50 | #include "undistort.simd_declarations.hpp" // defines CV_CPU_DISPATCH_MODES_ALL=AVX2,...,BASELINE based on CMakeLists.txt content |
51 | |
52 | namespace cv |
53 | { |
54 | |
55 | Mat getDefaultNewCameraMatrix( InputArray _cameraMatrix, Size imgsize, |
56 | bool centerPrincipalPoint ) |
57 | { |
58 | Mat cameraMatrix = _cameraMatrix.getMat(); |
59 | if( !centerPrincipalPoint && cameraMatrix.type() == CV_64F ) |
60 | return cameraMatrix; |
61 | |
62 | Mat newCameraMatrix; |
63 | cameraMatrix.convertTo(m: newCameraMatrix, CV_64F); |
64 | if( centerPrincipalPoint ) |
65 | { |
66 | newCameraMatrix.ptr<double>()[2] = (imgsize.width-1)*0.5; |
67 | newCameraMatrix.ptr<double>()[5] = (imgsize.height-1)*0.5; |
68 | } |
69 | return newCameraMatrix; |
70 | } |
71 | |
72 | namespace { |
73 | Ptr<ParallelLoopBody> getInitUndistortRectifyMapComputer(Size _size, Mat &_map1, Mat &_map2, int _m1type, |
74 | const double* _ir, Matx33d &_matTilt, |
75 | double _u0, double _v0, double _fx, double _fy, |
76 | double _k1, double _k2, double _p1, double _p2, |
77 | double _k3, double _k4, double _k5, double _k6, |
78 | double _s1, double _s2, double _s3, double _s4) |
79 | { |
80 | CV_INSTRUMENT_REGION(); |
81 | |
82 | CV_CPU_DISPATCH(getInitUndistortRectifyMapComputer, (_size, _map1, _map2, _m1type, _ir, _matTilt, _u0, _v0, _fx, _fy, _k1, _k2, _p1, _p2, _k3, _k4, _k5, _k6, _s1, _s2, _s3, _s4), |
83 | CV_CPU_DISPATCH_MODES_ALL); |
84 | } |
85 | } |
86 | |
87 | void initUndistortRectifyMap( InputArray _cameraMatrix, InputArray _distCoeffs, |
88 | InputArray _matR, InputArray _newCameraMatrix, |
89 | Size size, int m1type, OutputArray _map1, OutputArray _map2 ) |
90 | { |
91 | Mat cameraMatrix = _cameraMatrix.getMat(), distCoeffs = _distCoeffs.getMat(); |
92 | Mat matR = _matR.getMat(), newCameraMatrix = _newCameraMatrix.getMat(); |
93 | |
94 | if( m1type <= 0 ) |
95 | m1type = CV_16SC2; |
96 | CV_Assert( m1type == CV_16SC2 || m1type == CV_32FC1 || m1type == CV_32FC2 ); |
97 | _map1.create( sz: size, type: m1type ); |
98 | Mat map1 = _map1.getMat(), map2; |
99 | if( m1type != CV_32FC2 ) |
100 | { |
101 | _map2.create( sz: size, type: m1type == CV_16SC2 ? CV_16UC1 : CV_32FC1 ); |
102 | map2 = _map2.getMat(); |
103 | } |
104 | else |
105 | _map2.release(); |
106 | |
107 | Mat_<double> R = Mat_<double>::eye(rows: 3, cols: 3); |
108 | Mat_<double> A = Mat_<double>(cameraMatrix), Ar; |
109 | |
110 | if( !newCameraMatrix.empty() ) |
111 | Ar = Mat_<double>(newCameraMatrix); |
112 | else |
113 | Ar = getDefaultNewCameraMatrix( cameraMatrix: A, imgsize: size, centerPrincipalPoint: true ); |
114 | |
115 | if( !matR.empty() ) |
116 | R = Mat_<double>(matR); |
117 | |
118 | if( !distCoeffs.empty() ) |
119 | distCoeffs = Mat_<double>(distCoeffs); |
120 | else |
121 | { |
122 | distCoeffs.create(rows: 14, cols: 1, CV_64F); |
123 | distCoeffs = 0.; |
124 | } |
125 | |
126 | CV_Assert( A.size() == Size(3,3) && A.size() == R.size() ); |
127 | CV_Assert( Ar.size() == Size(3,3) || Ar.size() == Size(4, 3)); |
128 | Mat_<double> iR = (Ar.colRange(startcol: 0,endcol: 3)*R).inv(method: DECOMP_LU); |
129 | const double* ir = &iR(0,0); |
130 | |
131 | double u0 = A(0, 2), v0 = A(1, 2); |
132 | double fx = A(0, 0), fy = A(1, 1); |
133 | |
134 | CV_Assert( distCoeffs.size() == Size(1, 4) || distCoeffs.size() == Size(4, 1) || |
135 | distCoeffs.size() == Size(1, 5) || distCoeffs.size() == Size(5, 1) || |
136 | distCoeffs.size() == Size(1, 8) || distCoeffs.size() == Size(8, 1) || |
137 | distCoeffs.size() == Size(1, 12) || distCoeffs.size() == Size(12, 1) || |
138 | distCoeffs.size() == Size(1, 14) || distCoeffs.size() == Size(14, 1)); |
139 | |
140 | if( distCoeffs.rows != 1 && !distCoeffs.isContinuous() ) |
141 | distCoeffs = distCoeffs.t(); |
142 | |
143 | const double* const distPtr = distCoeffs.ptr<double>(); |
144 | double k1 = distPtr[0]; |
145 | double k2 = distPtr[1]; |
146 | double p1 = distPtr[2]; |
147 | double p2 = distPtr[3]; |
148 | double k3 = distCoeffs.cols + distCoeffs.rows - 1 >= 5 ? distPtr[4] : 0.; |
149 | double k4 = distCoeffs.cols + distCoeffs.rows - 1 >= 8 ? distPtr[5] : 0.; |
150 | double k5 = distCoeffs.cols + distCoeffs.rows - 1 >= 8 ? distPtr[6] : 0.; |
151 | double k6 = distCoeffs.cols + distCoeffs.rows - 1 >= 8 ? distPtr[7] : 0.; |
152 | double s1 = distCoeffs.cols + distCoeffs.rows - 1 >= 12 ? distPtr[8] : 0.; |
153 | double s2 = distCoeffs.cols + distCoeffs.rows - 1 >= 12 ? distPtr[9] : 0.; |
154 | double s3 = distCoeffs.cols + distCoeffs.rows - 1 >= 12 ? distPtr[10] : 0.; |
155 | double s4 = distCoeffs.cols + distCoeffs.rows - 1 >= 12 ? distPtr[11] : 0.; |
156 | double tauX = distCoeffs.cols + distCoeffs.rows - 1 >= 14 ? distPtr[12] : 0.; |
157 | double tauY = distCoeffs.cols + distCoeffs.rows - 1 >= 14 ? distPtr[13] : 0.; |
158 | |
159 | // Matrix for trapezoidal distortion of tilted image sensor |
160 | Matx33d matTilt = Matx33d::eye(); |
161 | detail::computeTiltProjectionMatrix(tauX, tauY, matTilt: &matTilt); |
162 | |
163 | parallel_for_(range: Range(0, size.height), body: *getInitUndistortRectifyMapComputer( |
164 | size: size, map1&: map1, map2&: map2, m1type: m1type, ir: ir, matTilt&: matTilt, u0: u0, v0: v0, |
165 | fx: fx, fy: fy, k1: k1, k2: k2, p1: p1, p2: p2, k3: k3, k4: k4, k5: k5, k6: k6, s1: s1, s2: s2, s3: s3, s4: s4)); |
166 | } |
167 | |
168 | void initInverseRectificationMap( InputArray _cameraMatrix, InputArray _distCoeffs, |
169 | InputArray _matR, InputArray _newCameraMatrix, |
170 | const Size& size, int m1type, OutputArray _map1, OutputArray _map2 ) |
171 | { |
172 | // Parameters |
173 | Mat cameraMatrix = _cameraMatrix.getMat(), distCoeffs = _distCoeffs.getMat(); |
174 | Mat matR = _matR.getMat(), newCameraMatrix = _newCameraMatrix.getMat(); |
175 | |
176 | // Check m1type validity |
177 | if( m1type <= 0 ) |
178 | m1type = CV_16SC2; |
179 | CV_Assert( m1type == CV_16SC2 || m1type == CV_32FC1 || m1type == CV_32FC2 ); |
180 | |
181 | // Init Maps |
182 | _map1.create( sz: size, type: m1type ); |
183 | Mat map1 = _map1.getMat(), map2; |
184 | if( m1type != CV_32FC2 ) |
185 | { |
186 | _map2.create( sz: size, type: m1type == CV_16SC2 ? CV_16UC1 : CV_32FC1 ); |
187 | map2 = _map2.getMat(); |
188 | } |
189 | else { |
190 | _map2.release(); |
191 | } |
192 | |
193 | // Init camera intrinsics |
194 | Mat_<double> A = Mat_<double>(cameraMatrix), Ar; |
195 | if( !newCameraMatrix.empty() ) |
196 | Ar = Mat_<double>(newCameraMatrix); |
197 | else |
198 | Ar = getDefaultNewCameraMatrix( cameraMatrix: A, imgsize: size, centerPrincipalPoint: true ); |
199 | CV_Assert( A.size() == Size(3,3) ); |
200 | CV_Assert( Ar.size() == Size(3,3) || Ar.size() == Size(4, 3)); |
201 | |
202 | // Init rotation matrix |
203 | Mat_<double> R = Mat_<double>::eye(rows: 3, cols: 3); |
204 | if( !matR.empty() ) |
205 | { |
206 | R = Mat_<double>(matR); |
207 | //Note, do not inverse |
208 | } |
209 | CV_Assert( Size(3,3) == R.size() ); |
210 | |
211 | // Init distortion vector |
212 | if( !distCoeffs.empty() ){ |
213 | distCoeffs = Mat_<double>(distCoeffs); |
214 | |
215 | // Fix distortion vector orientation |
216 | if( distCoeffs.rows != 1 && !distCoeffs.isContinuous() ) { |
217 | distCoeffs = distCoeffs.t(); |
218 | } |
219 | } |
220 | |
221 | // Validate distortion vector size |
222 | CV_Assert( distCoeffs.empty() || // Empty allows cv::undistortPoints to skip distortion |
223 | distCoeffs.size() == Size(1, 4) || distCoeffs.size() == Size(4, 1) || |
224 | distCoeffs.size() == Size(1, 5) || distCoeffs.size() == Size(5, 1) || |
225 | distCoeffs.size() == Size(1, 8) || distCoeffs.size() == Size(8, 1) || |
226 | distCoeffs.size() == Size(1, 12) || distCoeffs.size() == Size(12, 1) || |
227 | distCoeffs.size() == Size(1, 14) || distCoeffs.size() == Size(14, 1)); |
228 | |
229 | // Create objectPoints |
230 | std::vector<cv::Point2i> p2i_objPoints; |
231 | std::vector<cv::Point2f> p2f_objPoints; |
232 | for (int r = 0; r < size.height; r++) |
233 | { |
234 | for (int c = 0; c < size.width; c++) |
235 | { |
236 | p2i_objPoints.push_back(x: cv::Point2i(c, r)); |
237 | p2f_objPoints.push_back(x: cv::Point2f(static_cast<float>(c), static_cast<float>(r))); |
238 | } |
239 | } |
240 | |
241 | // Undistort |
242 | std::vector<cv::Point2f> p2f_objPoints_undistorted; |
243 | undistortPoints( |
244 | src: p2f_objPoints, |
245 | dst: p2f_objPoints_undistorted, |
246 | cameraMatrix: A, |
247 | distCoeffs, |
248 | R: cv::Mat::eye(size: cv::Size(3, 3), CV_64FC1), // R |
249 | P: cv::Mat::eye(size: cv::Size(3, 3), CV_64FC1) // P = New K |
250 | ); |
251 | |
252 | // Rectify |
253 | std::vector<cv::Point2f> p2f_sourcePoints_pinHole; |
254 | perspectiveTransform( |
255 | src: p2f_objPoints_undistorted, |
256 | dst: p2f_sourcePoints_pinHole, |
257 | m: R |
258 | ); |
259 | |
260 | // Project points back to camera coordinates. |
261 | std::vector<cv::Point2f> p2f_sourcePoints; |
262 | undistortPoints( |
263 | src: p2f_sourcePoints_pinHole, |
264 | dst: p2f_sourcePoints, |
265 | cameraMatrix: cv::Mat::eye(size: cv::Size(3, 3), CV_32FC1), // K |
266 | distCoeffs: cv::Mat::zeros(size: cv::Size(1, 4), CV_32FC1), // Distortion |
267 | R: cv::Mat::eye(size: cv::Size(3, 3), CV_32FC1), // R |
268 | P: Ar // New K |
269 | ); |
270 | |
271 | // Copy to map |
272 | if (m1type == CV_16SC2) { |
273 | for (size_t i=0; i < p2i_objPoints.size(); i++) { |
274 | map1.at<Vec2s>(i0: p2i_objPoints[i].y, i1: p2i_objPoints[i].x) = Vec2s(saturate_cast<short>(v: p2f_sourcePoints[i].x), saturate_cast<short>(v: p2f_sourcePoints[i].y)); |
275 | } |
276 | } else if (m1type == CV_32FC2) { |
277 | for (size_t i=0; i < p2i_objPoints.size(); i++) { |
278 | map1.at<Vec2f>(i0: p2i_objPoints[i].y, i1: p2i_objPoints[i].x) = Vec2f(p2f_sourcePoints[i]); |
279 | } |
280 | } else { // m1type == CV_32FC1 |
281 | for (size_t i=0; i < p2i_objPoints.size(); i++) { |
282 | map1.at<float>(i0: p2i_objPoints[i].y, i1: p2i_objPoints[i].x) = p2f_sourcePoints[i].x; |
283 | map2.at<float>(i0: p2i_objPoints[i].y, i1: p2i_objPoints[i].x) = p2f_sourcePoints[i].y; |
284 | } |
285 | } |
286 | } |
287 | |
288 | void undistort( InputArray _src, OutputArray _dst, InputArray _cameraMatrix, |
289 | InputArray _distCoeffs, InputArray _newCameraMatrix ) |
290 | { |
291 | CV_INSTRUMENT_REGION(); |
292 | |
293 | Mat src = _src.getMat(), cameraMatrix = _cameraMatrix.getMat(); |
294 | Mat distCoeffs = _distCoeffs.getMat(), newCameraMatrix = _newCameraMatrix.getMat(); |
295 | |
296 | _dst.create( sz: src.size(), type: src.type() ); |
297 | Mat dst = _dst.getMat(); |
298 | |
299 | CV_Assert( dst.data != src.data ); |
300 | |
301 | int stripe_size0 = std::min(a: std::max(a: 1, b: (1 << 12) / std::max(a: src.cols, b: 1)), b: src.rows); |
302 | Mat map1(stripe_size0, src.cols, CV_16SC2), map2(stripe_size0, src.cols, CV_16UC1); |
303 | |
304 | Mat_<double> A, Ar, I = Mat_<double>::eye(rows: 3,cols: 3); |
305 | |
306 | cameraMatrix.convertTo(m: A, CV_64F); |
307 | if( !distCoeffs.empty() ) |
308 | distCoeffs = Mat_<double>(distCoeffs); |
309 | else |
310 | { |
311 | distCoeffs.create(rows: 5, cols: 1, CV_64F); |
312 | distCoeffs = 0.; |
313 | } |
314 | |
315 | if( !newCameraMatrix.empty() ) |
316 | newCameraMatrix.convertTo(m: Ar, CV_64F); |
317 | else |
318 | A.copyTo(m: Ar); |
319 | |
320 | double v0 = Ar(1, 2); |
321 | for( int y = 0; y < src.rows; y += stripe_size0 ) |
322 | { |
323 | int stripe_size = std::min( a: stripe_size0, b: src.rows - y ); |
324 | Ar(1, 2) = v0 - y; |
325 | Mat map1_part = map1.rowRange(startrow: 0, endrow: stripe_size), |
326 | map2_part = map2.rowRange(startrow: 0, endrow: stripe_size), |
327 | dst_part = dst.rowRange(startrow: y, endrow: y + stripe_size); |
328 | |
329 | initUndistortRectifyMap( cameraMatrix: A, distCoeffs: distCoeffs, matR: I, newCameraMatrix: Ar, size: Size(src.cols, stripe_size), |
330 | m1type: map1_part.type(), map1: map1_part, map2: map2_part ); |
331 | remap( src, dst: dst_part, map1: map1_part, map2: map2_part, interpolation: INTER_LINEAR, borderMode: BORDER_CONSTANT ); |
332 | } |
333 | } |
334 | |
335 | } |
336 | |
337 | CV_IMPL void |
338 | cvUndistort2( const CvArr* srcarr, CvArr* dstarr, const CvMat* Aarr, const CvMat* dist_coeffs, const CvMat* newAarr ) |
339 | { |
340 | cv::Mat src = cv::cvarrToMat(arr: srcarr), dst = cv::cvarrToMat(arr: dstarr), dst0 = dst; |
341 | cv::Mat A = cv::cvarrToMat(arr: Aarr), distCoeffs = cv::cvarrToMat(arr: dist_coeffs), newA; |
342 | if( newAarr ) |
343 | newA = cv::cvarrToMat(arr: newAarr); |
344 | |
345 | CV_Assert( src.size() == dst.size() && src.type() == dst.type() ); |
346 | cv::undistort( src: src, dst: dst, cameraMatrix: A, distCoeffs: distCoeffs, newCameraMatrix: newA ); |
347 | } |
348 | |
349 | |
350 | CV_IMPL void cvInitUndistortMap( const CvMat* Aarr, const CvMat* dist_coeffs, |
351 | CvArr* mapxarr, CvArr* mapyarr ) |
352 | { |
353 | cv::Mat A = cv::cvarrToMat(arr: Aarr), distCoeffs = cv::cvarrToMat(arr: dist_coeffs); |
354 | cv::Mat mapx = cv::cvarrToMat(arr: mapxarr), mapy, mapx0 = mapx, mapy0; |
355 | |
356 | if( mapyarr ) |
357 | mapy0 = mapy = cv::cvarrToMat(arr: mapyarr); |
358 | |
359 | cv::initUndistortRectifyMap( cameraMatrix: A, distCoeffs: distCoeffs, matR: cv::Mat(), newCameraMatrix: A, |
360 | size: mapx.size(), m1type: mapx.type(), map1: mapx, map2: mapy ); |
361 | CV_Assert( mapx0.data == mapx.data && mapy0.data == mapy.data ); |
362 | } |
363 | |
364 | void |
365 | cvInitUndistortRectifyMap( const CvMat* Aarr, const CvMat* dist_coeffs, |
366 | const CvMat *Rarr, const CvMat* ArArr, CvArr* mapxarr, CvArr* mapyarr ) |
367 | { |
368 | cv::Mat A = cv::cvarrToMat(arr: Aarr), distCoeffs, R, Ar; |
369 | cv::Mat mapx = cv::cvarrToMat(arr: mapxarr), mapy, mapx0 = mapx, mapy0; |
370 | |
371 | if( mapyarr ) |
372 | mapy0 = mapy = cv::cvarrToMat(arr: mapyarr); |
373 | |
374 | if( dist_coeffs ) |
375 | distCoeffs = cv::cvarrToMat(arr: dist_coeffs); |
376 | if( Rarr ) |
377 | R = cv::cvarrToMat(arr: Rarr); |
378 | if( ArArr ) |
379 | Ar = cv::cvarrToMat(arr: ArArr); |
380 | |
381 | cv::initUndistortRectifyMap( cameraMatrix: A, distCoeffs: distCoeffs, matR: R, newCameraMatrix: Ar, size: mapx.size(), m1type: mapx.type(), map1: mapx, map2: mapy ); |
382 | CV_Assert( mapx0.data == mapx.data && mapy0.data == mapy.data ); |
383 | } |
384 | |
385 | static void cvUndistortPointsInternal( const CvMat* _src, CvMat* _dst, const CvMat* _cameraMatrix, |
386 | const CvMat* _distCoeffs, |
387 | const CvMat* matR, const CvMat* matP, cv::TermCriteria criteria) |
388 | { |
389 | CV_Assert(criteria.isValid()); |
390 | double A[3][3], RR[3][3], k[14]={0,0,0,0,0,0,0,0,0,0,0,0,0,0}; |
391 | CvMat matA=cvMat(rows: 3, cols: 3, CV_64F, data: A), _Dk; |
392 | CvMat _RR=cvMat(rows: 3, cols: 3, CV_64F, data: RR); |
393 | cv::Matx33d invMatTilt = cv::Matx33d::eye(); |
394 | cv::Matx33d matTilt = cv::Matx33d::eye(); |
395 | |
396 | CV_Assert( CV_IS_MAT(_src) && CV_IS_MAT(_dst) && |
397 | (_src->rows == 1 || _src->cols == 1) && |
398 | (_dst->rows == 1 || _dst->cols == 1) && |
399 | _src->cols + _src->rows - 1 == _dst->rows + _dst->cols - 1 && |
400 | (CV_MAT_TYPE(_src->type) == CV_32FC2 || CV_MAT_TYPE(_src->type) == CV_64FC2) && |
401 | (CV_MAT_TYPE(_dst->type) == CV_32FC2 || CV_MAT_TYPE(_dst->type) == CV_64FC2)); |
402 | |
403 | CV_Assert( CV_IS_MAT(_cameraMatrix) && |
404 | _cameraMatrix->rows == 3 && _cameraMatrix->cols == 3 ); |
405 | |
406 | cvConvert( _cameraMatrix, &matA ); |
407 | |
408 | |
409 | if( _distCoeffs ) |
410 | { |
411 | CV_Assert( CV_IS_MAT(_distCoeffs) && |
412 | (_distCoeffs->rows == 1 || _distCoeffs->cols == 1) && |
413 | (_distCoeffs->rows*_distCoeffs->cols == 4 || |
414 | _distCoeffs->rows*_distCoeffs->cols == 5 || |
415 | _distCoeffs->rows*_distCoeffs->cols == 8 || |
416 | _distCoeffs->rows*_distCoeffs->cols == 12 || |
417 | _distCoeffs->rows*_distCoeffs->cols == 14)); |
418 | |
419 | _Dk = cvMat( rows: _distCoeffs->rows, cols: _distCoeffs->cols, |
420 | CV_MAKETYPE(CV_64F,CV_MAT_CN(_distCoeffs->type)), data: k); |
421 | |
422 | cvConvert( _distCoeffs, &_Dk ); |
423 | if (k[12] != 0 || k[13] != 0) |
424 | { |
425 | cv::detail::computeTiltProjectionMatrix<double>(tauX: k[12], tauY: k[13], NULL, NULL, NULL, invMatTilt: &invMatTilt); |
426 | cv::detail::computeTiltProjectionMatrix<double>(tauX: k[12], tauY: k[13], matTilt: &matTilt, NULL, NULL); |
427 | } |
428 | } |
429 | |
430 | if( matR ) |
431 | { |
432 | CV_Assert( CV_IS_MAT(matR) && matR->rows == 3 && matR->cols == 3 ); |
433 | cvConvert( matR, &_RR ); |
434 | } |
435 | else |
436 | cvSetIdentity(mat: &_RR); |
437 | |
438 | if( matP ) |
439 | { |
440 | double PP[3][3]; |
441 | CvMat _P3x3, _PP=cvMat(rows: 3, cols: 3, CV_64F, data: PP); |
442 | CV_Assert( CV_IS_MAT(matP) && matP->rows == 3 && (matP->cols == 3 || matP->cols == 4)); |
443 | cvConvert( cvGetCols(matP, &_P3x3, 0, 3), &_PP ); |
444 | cvMatMul( &_PP, &_RR, &_RR ); |
445 | } |
446 | |
447 | const CvPoint2D32f* srcf = (const CvPoint2D32f*)_src->data.ptr; |
448 | const CvPoint2D64f* srcd = (const CvPoint2D64f*)_src->data.ptr; |
449 | CvPoint2D32f* dstf = (CvPoint2D32f*)_dst->data.ptr; |
450 | CvPoint2D64f* dstd = (CvPoint2D64f*)_dst->data.ptr; |
451 | int stype = CV_MAT_TYPE(_src->type); |
452 | int dtype = CV_MAT_TYPE(_dst->type); |
453 | int sstep = _src->rows == 1 ? 1 : _src->step/CV_ELEM_SIZE(stype); |
454 | int dstep = _dst->rows == 1 ? 1 : _dst->step/CV_ELEM_SIZE(dtype); |
455 | |
456 | double fx = A[0][0]; |
457 | double fy = A[1][1]; |
458 | double ifx = 1./fx; |
459 | double ify = 1./fy; |
460 | double cx = A[0][2]; |
461 | double cy = A[1][2]; |
462 | |
463 | int n = _src->rows + _src->cols - 1; |
464 | for( int i = 0; i < n; i++ ) |
465 | { |
466 | double x, y, x0 = 0, y0 = 0, u, v; |
467 | if( stype == CV_32FC2 ) |
468 | { |
469 | x = srcf[i*sstep].x; |
470 | y = srcf[i*sstep].y; |
471 | } |
472 | else |
473 | { |
474 | x = srcd[i*sstep].x; |
475 | y = srcd[i*sstep].y; |
476 | } |
477 | u = x; v = y; |
478 | x = (x - cx)*ifx; |
479 | y = (y - cy)*ify; |
480 | |
481 | if( _distCoeffs ) { |
482 | // compensate tilt distortion |
483 | cv::Vec3d vecUntilt = invMatTilt * cv::Vec3d(x, y, 1); |
484 | double invProj = vecUntilt(2) ? 1./vecUntilt(2) : 1; |
485 | x0 = x = invProj * vecUntilt(0); |
486 | y0 = y = invProj * vecUntilt(1); |
487 | |
488 | double error = std::numeric_limits<double>::max(); |
489 | // compensate distortion iteratively |
490 | |
491 | for( int j = 0; ; j++ ) |
492 | { |
493 | if ((criteria.type & cv::TermCriteria::COUNT) && j >= criteria.maxCount) |
494 | break; |
495 | if ((criteria.type & cv::TermCriteria::EPS) && error < criteria.epsilon) |
496 | break; |
497 | double r2 = x*x + y*y; |
498 | double icdist = (1 + ((k[7]*r2 + k[6])*r2 + k[5])*r2)/(1 + ((k[4]*r2 + k[1])*r2 + k[0])*r2); |
499 | if (icdist < 0) // test: undistortPoints.regression_14583 |
500 | { |
501 | x = (u - cx)*ifx; |
502 | y = (v - cy)*ify; |
503 | break; |
504 | } |
505 | double deltaX = 2*k[2]*x*y + k[3]*(r2 + 2*x*x)+ k[8]*r2+k[9]*r2*r2; |
506 | double deltaY = k[2]*(r2 + 2*y*y) + 2*k[3]*x*y+ k[10]*r2+k[11]*r2*r2; |
507 | x = (x0 - deltaX)*icdist; |
508 | y = (y0 - deltaY)*icdist; |
509 | |
510 | if(criteria.type & cv::TermCriteria::EPS) |
511 | { |
512 | double r4, r6, a1, a2, a3, cdist, icdist2; |
513 | double xd, yd, xd0, yd0; |
514 | cv::Vec3d vecTilt; |
515 | |
516 | r2 = x*x + y*y; |
517 | r4 = r2*r2; |
518 | r6 = r4*r2; |
519 | a1 = 2*x*y; |
520 | a2 = r2 + 2*x*x; |
521 | a3 = r2 + 2*y*y; |
522 | cdist = 1 + k[0]*r2 + k[1]*r4 + k[4]*r6; |
523 | icdist2 = 1./(1 + k[5]*r2 + k[6]*r4 + k[7]*r6); |
524 | xd0 = x*cdist*icdist2 + k[2]*a1 + k[3]*a2 + k[8]*r2+k[9]*r4; |
525 | yd0 = y*cdist*icdist2 + k[2]*a3 + k[3]*a1 + k[10]*r2+k[11]*r4; |
526 | |
527 | vecTilt = matTilt*cv::Vec3d(xd0, yd0, 1); |
528 | invProj = vecTilt(2) ? 1./vecTilt(2) : 1; |
529 | xd = invProj * vecTilt(0); |
530 | yd = invProj * vecTilt(1); |
531 | |
532 | double x_proj = xd*fx + cx; |
533 | double y_proj = yd*fy + cy; |
534 | |
535 | error = sqrt( x: pow(x: x_proj - u, y: 2) + pow(x: y_proj - v, y: 2) ); |
536 | } |
537 | } |
538 | } |
539 | |
540 | double xx = RR[0][0]*x + RR[0][1]*y + RR[0][2]; |
541 | double yy = RR[1][0]*x + RR[1][1]*y + RR[1][2]; |
542 | double ww = 1./(RR[2][0]*x + RR[2][1]*y + RR[2][2]); |
543 | x = xx*ww; |
544 | y = yy*ww; |
545 | |
546 | if( dtype == CV_32FC2 ) |
547 | { |
548 | dstf[i*dstep].x = (float)x; |
549 | dstf[i*dstep].y = (float)y; |
550 | } |
551 | else |
552 | { |
553 | dstd[i*dstep].x = x; |
554 | dstd[i*dstep].y = y; |
555 | } |
556 | } |
557 | } |
558 | |
559 | void cvUndistortPoints(const CvMat* _src, CvMat* _dst, const CvMat* _cameraMatrix, |
560 | const CvMat* _distCoeffs, |
561 | const CvMat* matR, const CvMat* matP) |
562 | { |
563 | cvUndistortPointsInternal(_src, _dst, _cameraMatrix, _distCoeffs, matR, matP, |
564 | criteria: cv::TermCriteria(cv::TermCriteria::COUNT, 5, 0.01)); |
565 | } |
566 | |
567 | namespace cv { |
568 | |
569 | void undistortPoints(InputArray _src, OutputArray _dst, |
570 | InputArray _cameraMatrix, |
571 | InputArray _distCoeffs, |
572 | InputArray _Rmat, |
573 | InputArray _Pmat) |
574 | { |
575 | undistortPoints(src: _src, dst: _dst, cameraMatrix: _cameraMatrix, distCoeffs: _distCoeffs, R: _Rmat, P: _Pmat, criteria: TermCriteria(TermCriteria::MAX_ITER, 5, 0.01)); |
576 | } |
577 | |
578 | void undistortPoints(InputArray _src, OutputArray _dst, |
579 | InputArray _cameraMatrix, |
580 | InputArray _distCoeffs, |
581 | InputArray _Rmat, |
582 | InputArray _Pmat, |
583 | TermCriteria criteria) |
584 | { |
585 | Mat src = _src.getMat(), cameraMatrix = _cameraMatrix.getMat(); |
586 | Mat distCoeffs = _distCoeffs.getMat(), R = _Rmat.getMat(), P = _Pmat.getMat(); |
587 | |
588 | int npoints = src.checkVector(elemChannels: 2), depth = src.depth(); |
589 | if (npoints < 0) |
590 | src = src.t(); |
591 | npoints = src.checkVector(elemChannels: 2); |
592 | CV_Assert(npoints >= 0 && src.isContinuous() && (depth == CV_32F || depth == CV_64F)); |
593 | |
594 | if (src.cols == 2) |
595 | src = src.reshape(cn: 2); |
596 | |
597 | _dst.create(rows: npoints, cols: 1, CV_MAKETYPE(depth, 2), i: -1, allowTransposed: true); |
598 | Mat dst = _dst.getMat(); |
599 | |
600 | CvMat _csrc = cvMat(m: src), _cdst = cvMat(m: dst), _ccameraMatrix = cvMat(m: cameraMatrix); |
601 | CvMat matR, matP, _cdistCoeffs, *pR=0, *pP=0, *pD=0; |
602 | if( !R.empty() ) |
603 | pR = &(matR = cvMat(m: R)); |
604 | if( !P.empty() ) |
605 | pP = &(matP = cvMat(m: P)); |
606 | if( !distCoeffs.empty() ) |
607 | pD = &(_cdistCoeffs = cvMat(m: distCoeffs)); |
608 | cvUndistortPointsInternal(src: &_csrc, dst: &_cdst, cameraMatrix: &_ccameraMatrix, distCoeffs: pD, matR: pR, matP: pP, criteria); |
609 | } |
610 | |
611 | void undistortImagePoints(InputArray src, OutputArray dst, InputArray cameraMatrix, InputArray distCoeffs, TermCriteria termCriteria) |
612 | { |
613 | undistortPoints(src: src, dst: dst, cameraMatrix: cameraMatrix, distCoeffs: distCoeffs, Rmat: noArray(), Pmat: cameraMatrix, criteria: termCriteria); |
614 | } |
615 | |
616 | static Point2f mapPointSpherical(const Point2f& p, float alpha, Vec4d* J, enum UndistortTypes projType) |
617 | { |
618 | double x = p.x, y = p.y; |
619 | double beta = 1 + 2*alpha; |
620 | double v = x*x + y*y + 1, iv = 1/v; |
621 | double u = std::sqrt(x: beta*v + alpha*alpha); |
622 | |
623 | double k = (u - alpha)*iv; |
624 | double kv = (v*beta/u - (u - alpha)*2)*iv*iv; |
625 | double kx = kv*x, ky = kv*y; |
626 | |
627 | if( projType == PROJ_SPHERICAL_ORTHO ) |
628 | { |
629 | if(J) |
630 | *J = Vec4d(kx*x + k, kx*y, ky*x, ky*y + k); |
631 | return Point2f((float)(x*k), (float)(y*k)); |
632 | } |
633 | if( projType == PROJ_SPHERICAL_EQRECT ) |
634 | { |
635 | // equirectangular |
636 | double iR = 1/(alpha + 1); |
637 | double x1 = std::max(a: std::min(a: x*k*iR, b: 1.), b: -1.); |
638 | double y1 = std::max(a: std::min(a: y*k*iR, b: 1.), b: -1.); |
639 | |
640 | if(J) |
641 | { |
642 | double fx1 = iR/std::sqrt(x: 1 - x1*x1); |
643 | double fy1 = iR/std::sqrt(x: 1 - y1*y1); |
644 | *J = Vec4d(fx1*(kx*x + k), fx1*ky*x, fy1*kx*y, fy1*(ky*y + k)); |
645 | } |
646 | return Point2f((float)asin(x: x1), (float)asin(x: y1)); |
647 | } |
648 | CV_Error(Error::StsBadArg, "Unknown projection type"); |
649 | } |
650 | |
651 | |
652 | static Point2f invMapPointSpherical(Point2f _p, float alpha, enum UndistortTypes projType) |
653 | { |
654 | double eps = 1e-12; |
655 | Vec2d p(_p.x, _p.y), q(_p.x, _p.y), err; |
656 | Vec4d J; |
657 | int i, maxiter = 5; |
658 | |
659 | for( i = 0; i < maxiter; i++ ) |
660 | { |
661 | Point2f p1 = mapPointSpherical(p: Point2f((float)q[0], (float)q[1]), alpha, J: &J, projType); |
662 | err = Vec2d(p1.x, p1.y) - p; |
663 | if( err[0]*err[0] + err[1]*err[1] < eps ) |
664 | break; |
665 | |
666 | Vec4d JtJ(J[0]*J[0] + J[2]*J[2], J[0]*J[1] + J[2]*J[3], |
667 | J[0]*J[1] + J[2]*J[3], J[1]*J[1] + J[3]*J[3]); |
668 | double d = JtJ[0]*JtJ[3] - JtJ[1]*JtJ[2]; |
669 | d = d ? 1./d : 0; |
670 | Vec4d iJtJ(JtJ[3]*d, -JtJ[1]*d, -JtJ[2]*d, JtJ[0]*d); |
671 | Vec2d JtErr(J[0]*err[0] + J[2]*err[1], J[1]*err[0] + J[3]*err[1]); |
672 | |
673 | q -= Vec2d(iJtJ[0]*JtErr[0] + iJtJ[1]*JtErr[1], iJtJ[2]*JtErr[0] + iJtJ[3]*JtErr[1]); |
674 | //Matx22d J(kx*x + k, kx*y, ky*x, ky*y + k); |
675 | //q -= Vec2d((J.t()*J).inv()*(J.t()*err)); |
676 | } |
677 | |
678 | return i < maxiter ? Point2f((float)q[0], (float)q[1]) : Point2f(-FLT_MAX, -FLT_MAX); |
679 | } |
680 | |
681 | float initWideAngleProjMap(InputArray _cameraMatrix0, InputArray _distCoeffs0, |
682 | Size imageSize, int destImageWidth, int m1type, |
683 | OutputArray _map1, OutputArray _map2, |
684 | enum UndistortTypes projType, double _alpha) |
685 | { |
686 | Mat cameraMatrix0 = _cameraMatrix0.getMat(), distCoeffs0 = _distCoeffs0.getMat(); |
687 | double k[14] = {0,0,0,0,0,0,0,0,0,0,0,0,0,0}, M[9]={0,0,0,0,0,0,0,0,0}; |
688 | Mat distCoeffs(distCoeffs0.rows, distCoeffs0.cols, CV_MAKETYPE(CV_64F,distCoeffs0.channels()), k); |
689 | Mat cameraMatrix(3,3,CV_64F,M); |
690 | Point2f scenter((float)cameraMatrix.at<double>(i0: 0,i1: 2), (float)cameraMatrix.at<double>(i0: 1,i1: 2)); |
691 | Point2f dcenter((destImageWidth-1)*0.5f, 0.f); |
692 | float xmin = FLT_MAX, xmax = -FLT_MAX, ymin = FLT_MAX, ymax = -FLT_MAX; |
693 | int N = 9; |
694 | std::vector<Point2f> uvec(1), vvec(1); |
695 | Mat I = Mat::eye(rows: 3,cols: 3,CV_64F); |
696 | float alpha = (float)_alpha; |
697 | |
698 | int ndcoeffs = distCoeffs0.cols*distCoeffs0.rows*distCoeffs0.channels(); |
699 | CV_Assert((distCoeffs0.cols == 1 || distCoeffs0.rows == 1) && |
700 | (ndcoeffs == 4 || ndcoeffs == 5 || ndcoeffs == 8 || ndcoeffs == 12 || ndcoeffs == 14)); |
701 | CV_Assert(cameraMatrix0.size() == Size(3,3)); |
702 | distCoeffs0.convertTo(m: distCoeffs,CV_64F); |
703 | cameraMatrix0.convertTo(m: cameraMatrix,CV_64F); |
704 | |
705 | alpha = std::min(a: alpha, b: 0.999f); |
706 | |
707 | for( int i = 0; i < N; i++ ) |
708 | for( int j = 0; j < N; j++ ) |
709 | { |
710 | Point2f p((float)j*imageSize.width/(N-1), (float)i*imageSize.height/(N-1)); |
711 | uvec[0] = p; |
712 | undistortPoints(src: uvec, dst: vvec, cameraMatrix: cameraMatrix, distCoeffs: distCoeffs, Rmat: I, Pmat: I); |
713 | Point2f q = mapPointSpherical(p: vvec[0], alpha, J: 0, projType); |
714 | if( xmin > q.x ) xmin = q.x; |
715 | if( xmax < q.x ) xmax = q.x; |
716 | if( ymin > q.y ) ymin = q.y; |
717 | if( ymax < q.y ) ymax = q.y; |
718 | } |
719 | |
720 | float scale = (float)std::min(a: dcenter.x/fabs(x: xmax), b: dcenter.x/fabs(x: xmin)); |
721 | Size dsize(destImageWidth, cvCeil(value: std::max(a: scale*fabs(x: ymin)*2, b: scale*fabs(x: ymax)*2))); |
722 | dcenter.y = (dsize.height - 1)*0.5f; |
723 | |
724 | Mat mapxy(dsize, CV_32FC2); |
725 | double k1 = k[0], k2 = k[1], k3 = k[2], p1 = k[3], p2 = k[4], k4 = k[5], k5 = k[6], k6 = k[7], s1 = k[8], s2 = k[9], s3 = k[10], s4 = k[11]; |
726 | double fx = cameraMatrix.at<double>(i0: 0,i1: 0), fy = cameraMatrix.at<double>(i0: 1,i1: 1), cx = scenter.x, cy = scenter.y; |
727 | cv::Matx33d matTilt; |
728 | cv::detail::computeTiltProjectionMatrix(tauX: k[12], tauY: k[13], matTilt: &matTilt); |
729 | |
730 | for( int y = 0; y < dsize.height; y++ ) |
731 | { |
732 | Point2f* mxy = mapxy.ptr<Point2f>(y); |
733 | for( int x = 0; x < dsize.width; x++ ) |
734 | { |
735 | Point2f p = (Point2f((float)x, (float)y) - dcenter)*(1.f/scale); |
736 | Point2f q = invMapPointSpherical(p: p, alpha, projType); |
737 | if( q.x <= -FLT_MAX && q.y <= -FLT_MAX ) |
738 | { |
739 | mxy[x] = Point2f(-1.f, -1.f); |
740 | continue; |
741 | } |
742 | double x2 = q.x*q.x, y2 = q.y*q.y; |
743 | double r2 = x2 + y2, _2xy = 2*q.x*q.y; |
744 | double kr = 1 + ((k3*r2 + k2)*r2 + k1)*r2/(1 + ((k6*r2 + k5)*r2 + k4)*r2); |
745 | double xd = (q.x*kr + p1*_2xy + p2*(r2 + 2*x2) + s1*r2+ s2*r2*r2); |
746 | double yd = (q.y*kr + p1*(r2 + 2*y2) + p2*_2xy + s3*r2+ s4*r2*r2); |
747 | cv::Vec3d vecTilt = matTilt*cv::Vec3d(xd, yd, 1); |
748 | double invProj = vecTilt(2) ? 1./vecTilt(2) : 1; |
749 | double u = fx*invProj*vecTilt(0) + cx; |
750 | double v = fy*invProj*vecTilt(1) + cy; |
751 | |
752 | mxy[x] = Point2f((float)u, (float)v); |
753 | } |
754 | } |
755 | |
756 | if(m1type == CV_32FC2) |
757 | { |
758 | _map1.create(sz: mapxy.size(), type: mapxy.type()); |
759 | Mat map1 = _map1.getMat(); |
760 | mapxy.copyTo(m: map1); |
761 | _map2.release(); |
762 | } |
763 | else |
764 | convertMaps(map1: mapxy, map2: Mat(), dstmap1: _map1, dstmap2: _map2, dstmap1type: m1type, nninterpolation: false); |
765 | |
766 | return scale; |
767 | } |
768 | |
769 | } // namespace |
770 | /* End of file */ |
771 |
Definitions
- getDefaultNewCameraMatrix
- getInitUndistortRectifyMapComputer
- initUndistortRectifyMap
- initInverseRectificationMap
- undistort
- cvUndistort2
- cvInitUndistortMap
- cvInitUndistortRectifyMap
- cvUndistortPointsInternal
- cvUndistortPoints
- undistortPoints
- undistortPoints
- undistortImagePoints
- mapPointSpherical
- invMapPointSpherical
Learn to use CMake with our Intro Training
Find out more