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.
10 // Intel License Agreement
11 // For Open Source Computer Vision Library
13 // Copyright (C) 2000, Intel Corporation, all rights reserved.
14 // Third party copyrights are property of their respective owners.
16 // Redistribution and use in source and binary forms, with or without modification,
17 // are permitted provided that the following conditions are met:
19 // * Redistribution's of source code must retain the above copyright notice,
20 // this list of conditions and the following disclaimer.
22 // * Redistribution's in binary form must reproduce the above copyright notice,
23 // this list of conditions and the following disclaimer in the documentation
24 // and/or other materials provided with the distribution.
26 // * The name of Intel Corporation may not be used to endorse or promote products
27 // derived from this software without specific prior written permission.
29 // This software is provided by the copyright holders and contributors "as is" and
30 // any express or implied warranties, including, but not limited to, the implied
31 // warranties of merchantability and fitness for a particular purpose are disclaimed.
32 // In no event shall the Intel Corporation or contributors be liable for any direct,
33 // indirect, incidental, special, exemplary, or consequential damages
34 // (including, but not limited to, procurement of substitute goods or services;
35 // loss of use, data, or profits; or business interruption) however caused
36 // and on any theory of liability, whether in contract, strict liability,
37 // or tort (including negligence or otherwise) arising in any way out of
38 // the use of this software, even if advised of the possibility of such damage.
42 #include "precomp.hpp"
43 #include "_modelest.h"
47 template<typename T> int icvCompressPoints( T* ptr, const uchar* mask, int mstep, int count )
50 for( i = j = 0; i < count; i++ )
60 class CvHomographyEstimator : public CvModelEstimator2
63 CvHomographyEstimator( int modelPoints );
65 virtual int runKernel( const CvMat* m1, const CvMat* m2, CvMat* model );
66 virtual bool refine( const CvMat* m1, const CvMat* m2,
67 CvMat* model, int maxIters );
69 virtual void computeReprojError( const CvMat* m1, const CvMat* m2,
70 const CvMat* model, CvMat* error );
74 CvHomographyEstimator::CvHomographyEstimator(int _modelPoints)
75 : CvModelEstimator2(_modelPoints, cvSize(3,3), 1)
77 assert( _modelPoints == 4 || _modelPoints == 5 );
78 checkPartialSubsets = false;
81 int CvHomographyEstimator::runKernel( const CvMat* m1, const CvMat* m2, CvMat* H )
83 int i, count = m1->rows*m1->cols;
84 const CvPoint2D64f* M = (const CvPoint2D64f*)m1->data.ptr;
85 const CvPoint2D64f* m = (const CvPoint2D64f*)m2->data.ptr;
87 double LtL[9][9], W[9][1], V[9][9];
88 CvMat _LtL = cvMat( 9, 9, CV_64F, LtL );
89 CvMat matW = cvMat( 9, 1, CV_64F, W );
90 CvMat matV = cvMat( 9, 9, CV_64F, V );
91 CvMat _H0 = cvMat( 3, 3, CV_64F, V[8] );
92 CvMat _Htemp = cvMat( 3, 3, CV_64F, V[7] );
93 CvPoint2D64f cM={0,0}, cm={0,0}, sM={0,0}, sm={0,0};
95 for( i = 0; i < count; i++ )
97 cm.x += m[i].x; cm.y += m[i].y;
98 cM.x += M[i].x; cM.y += M[i].y;
101 cm.x /= count; cm.y /= count;
102 cM.x /= count; cM.y /= count;
104 for( i = 0; i < count; i++ )
106 sm.x += fabs(m[i].x - cm.x);
107 sm.y += fabs(m[i].y - cm.y);
108 sM.x += fabs(M[i].x - cM.x);
109 sM.y += fabs(M[i].y - cM.y);
112 if( fabs(sm.x) < DBL_EPSILON || fabs(sm.y) < DBL_EPSILON ||
113 fabs(sM.x) < DBL_EPSILON || fabs(sM.y) < DBL_EPSILON )
115 sm.x = count/sm.x; sm.y = count/sm.y;
116 sM.x = count/sM.x; sM.y = count/sM.y;
118 double invHnorm[9] = { 1./sm.x, 0, cm.x, 0, 1./sm.y, cm.y, 0, 0, 1 };
119 double Hnorm2[9] = { sM.x, 0, -cM.x*sM.x, 0, sM.y, -cM.y*sM.y, 0, 0, 1 };
120 CvMat _invHnorm = cvMat( 3, 3, CV_64FC1, invHnorm );
121 CvMat _Hnorm2 = cvMat( 3, 3, CV_64FC1, Hnorm2 );
124 for( i = 0; i < count; i++ )
126 double x = (m[i].x - cm.x)*sm.x, y = (m[i].y - cm.y)*sm.y;
127 double X = (M[i].x - cM.x)*sM.x, Y = (M[i].y - cM.y)*sM.y;
128 double Lx[] = { X, Y, 1, 0, 0, 0, -x*X, -x*Y, -x };
129 double Ly[] = { 0, 0, 0, X, Y, 1, -y*X, -y*Y, -y };
131 for( j = 0; j < 9; j++ )
132 for( k = j; k < 9; k++ )
133 LtL[j][k] += Lx[j]*Lx[k] + Ly[j]*Ly[k];
135 cvCompleteSymm( &_LtL );
137 //cvSVD( &_LtL, &matW, 0, &matV, CV_SVD_MODIFY_A + CV_SVD_V_T );
138 cvEigenVV( &_LtL, &matV, &matW );
139 cvMatMul( &_invHnorm, &_H0, &_Htemp );
140 cvMatMul( &_Htemp, &_Hnorm2, &_H0 );
141 cvConvertScale( &_H0, H, 1./_H0.data.db[8] );
147 void CvHomographyEstimator::computeReprojError( const CvMat* m1, const CvMat* m2,
148 const CvMat* model, CvMat* _err )
150 int i, count = m1->rows*m1->cols;
151 const CvPoint2D64f* M = (const CvPoint2D64f*)m1->data.ptr;
152 const CvPoint2D64f* m = (const CvPoint2D64f*)m2->data.ptr;
153 const double* H = model->data.db;
154 float* err = _err->data.fl;
156 for( i = 0; i < count; i++ )
158 double ww = 1./(H[6]*M[i].x + H[7]*M[i].y + 1.);
159 double dx = (H[0]*M[i].x + H[1]*M[i].y + H[2])*ww - m[i].x;
160 double dy = (H[3]*M[i].x + H[4]*M[i].y + H[5])*ww - m[i].y;
161 err[i] = (float)(dx*dx + dy*dy);
165 bool CvHomographyEstimator::refine( const CvMat* m1, const CvMat* m2, CvMat* model, int maxIters )
167 CvLevMarq solver(8, 0, cvTermCriteria(CV_TERMCRIT_ITER+CV_TERMCRIT_EPS, maxIters, DBL_EPSILON));
168 int i, j, k, count = m1->rows*m1->cols;
169 const CvPoint2D64f* M = (const CvPoint2D64f*)m1->data.ptr;
170 const CvPoint2D64f* m = (const CvPoint2D64f*)m2->data.ptr;
171 CvMat modelPart = cvMat( solver.param->rows, solver.param->cols, model->type, model->data.ptr );
172 cvCopy( &modelPart, solver.param );
176 const CvMat* _param = 0;
177 CvMat *_JtJ = 0, *_JtErr = 0;
178 double* _errNorm = 0;
180 if( !solver.updateAlt( _param, _JtJ, _JtErr, _errNorm ))
183 for( i = 0; i < count; i++ )
185 const double* h = _param->data.db;
186 double Mx = M[i].x, My = M[i].y;
187 double ww = h[6]*Mx + h[7]*My + 1.;
188 ww = fabs(ww) > DBL_EPSILON ? 1./ww : 0;
189 double _xi = (h[0]*Mx + h[1]*My + h[2])*ww;
190 double _yi = (h[3]*Mx + h[4]*My + h[5])*ww;
191 double err[] = { _xi - m[i].x, _yi - m[i].y };
196 { Mx*ww, My*ww, ww, 0, 0, 0, -Mx*ww*_xi, -My*ww*_xi },
197 { 0, 0, 0, Mx*ww, My*ww, ww, -Mx*ww*_yi, -My*ww*_yi }
200 for( j = 0; j < 8; j++ )
202 for( k = j; k < 8; k++ )
203 _JtJ->data.db[j*8+k] += J[0][j]*J[0][k] + J[1][j]*J[1][k];
204 _JtErr->data.db[j] += J[0][j]*err[0] + J[1][j]*err[1];
208 *_errNorm += err[0]*err[0] + err[1]*err[1];
212 cvCopy( solver.param, &modelPart );
218 cvFindHomography( const CvMat* objectPoints, const CvMat* imagePoints,
219 CvMat* __H, int method, double ransacReprojThreshold,
222 const double confidence = 0.995;
223 const int maxIters = 2000;
224 const double defaultRANSACReprojThreshold = 3;
226 Ptr<CvMat> m, M, tempMask;
229 CvMat matH = cvMat( 3, 3, CV_64FC1, H );
232 CV_Assert( CV_IS_MAT(imagePoints) && CV_IS_MAT(objectPoints) );
234 count = MAX(imagePoints->cols, imagePoints->rows);
235 CV_Assert( count >= 4 );
236 if( ransacReprojThreshold <= 0 )
237 ransacReprojThreshold = defaultRANSACReprojThreshold;
239 m = cvCreateMat( 1, count, CV_64FC2 );
240 cvConvertPointsHomogeneous( imagePoints, m );
242 M = cvCreateMat( 1, count, CV_64FC2 );
243 cvConvertPointsHomogeneous( objectPoints, M );
247 CV_Assert( CV_IS_MASK_ARR(mask) && CV_IS_MAT_CONT(mask->type) &&
248 (mask->rows == 1 || mask->cols == 1) &&
249 mask->rows*mask->cols == count );
251 if( mask || count > 4 )
252 tempMask = cvCreateMat( 1, count, CV_8U );
253 if( !tempMask.empty() )
254 cvSet( tempMask, cvScalarAll(1.) );
256 CvHomographyEstimator estimator(4);
259 if( method == CV_LMEDS )
260 result = estimator.runLMeDS( M, m, &matH, tempMask, confidence, maxIters );
261 else if( method == CV_RANSAC )
262 result = estimator.runRANSAC( M, m, &matH, tempMask, ransacReprojThreshold, confidence, maxIters);
264 result = estimator.runKernel( M, m, &matH ) > 0;
266 if( result && count > 4 )
268 icvCompressPoints( (CvPoint2D64f*)M->data.ptr, tempMask->data.ptr, 1, count );
269 count = icvCompressPoints( (CvPoint2D64f*)m->data.ptr, tempMask->data.ptr, 1, count );
270 M->cols = m->cols = count;
271 if( method == CV_RANSAC )
272 estimator.runKernel( M, m, &matH );
273 estimator.refine( M, m, &matH, 10 );
277 cvConvert( &matH, __H );
279 if( mask && tempMask )
281 if( CV_ARE_SIZES_EQ(mask, tempMask) )
282 cvCopy( tempMask, mask );
284 cvTranspose( tempMask, mask );
291 /* Evaluation of Fundamental Matrix from point correspondences.
292 The original code has been written by Valery Mosyagin */
294 /* The algorithms (except for RANSAC) and the notation have been taken from
295 Zhengyou Zhang's research report
296 "Determining the Epipolar Geometry and its Uncertainty: A Review"
297 that can be found at http://www-sop.inria.fr/robotvis/personnel/zzhang/zzhang-eng.html */
299 /************************************** 7-point algorithm *******************************/
300 class CvFMEstimator : public CvModelEstimator2
303 CvFMEstimator( int _modelPoints );
305 virtual int runKernel( const CvMat* m1, const CvMat* m2, CvMat* model );
306 virtual int run7Point( const CvMat* m1, const CvMat* m2, CvMat* model );
307 virtual int run8Point( const CvMat* m1, const CvMat* m2, CvMat* model );
309 virtual void computeReprojError( const CvMat* m1, const CvMat* m2,
310 const CvMat* model, CvMat* error );
313 CvFMEstimator::CvFMEstimator( int _modelPoints )
314 : CvModelEstimator2( _modelPoints, cvSize(3,3), _modelPoints == 7 ? 3 : 1 )
316 assert( _modelPoints == 7 || _modelPoints == 8 );
320 int CvFMEstimator::runKernel( const CvMat* m1, const CvMat* m2, CvMat* model )
322 return modelPoints == 7 ? run7Point( m1, m2, model ) : run8Point( m1, m2, model );
325 int CvFMEstimator::run7Point( const CvMat* _m1, const CvMat* _m2, CvMat* _fmatrix )
327 double a[7*9], w[7], v[9*9], c[4], r[3];
330 CvMat A = cvMat( 7, 9, CV_64F, a );
331 CvMat V = cvMat( 9, 9, CV_64F, v );
332 CvMat W = cvMat( 7, 1, CV_64F, w );
333 CvMat coeffs = cvMat( 1, 4, CV_64F, c );
334 CvMat roots = cvMat( 1, 3, CV_64F, r );
335 const CvPoint2D64f* m1 = (const CvPoint2D64f*)_m1->data.ptr;
336 const CvPoint2D64f* m2 = (const CvPoint2D64f*)_m2->data.ptr;
337 double* fmatrix = _fmatrix->data.db;
340 // form a linear system: i-th row of A(=a) represents
341 // the equation: (m2[i], 1)'*F*(m1[i], 1) = 0
342 for( i = 0; i < 7; i++ )
344 double x0 = m1[i].x, y0 = m1[i].y;
345 double x1 = m2[i].x, y1 = m2[i].y;
358 // A*(f11 f12 ... f33)' = 0 is singular (7 equations for 9 variables), so
359 // the solution is linear subspace of dimensionality 2.
360 // => use the last two singular vectors as a basis of the space
361 // (according to SVD properties)
362 cvSVD( &A, &W, 0, &V, CV_SVD_MODIFY_A + CV_SVD_V_T );
366 // f1, f2 is a basis => lambda*f1 + mu*f2 is an arbitrary f. matrix.
367 // as it is determined up to a scale, normalize lambda & mu (lambda + mu = 1),
368 // so f ~ lambda*f1 + (1 - lambda)*f2.
369 // use the additional constraint det(f) = det(lambda*f1 + (1-lambda)*f2) to find lambda.
370 // it will be a cubic equation.
371 // find c - polynomial coefficients.
372 for( i = 0; i < 9; i++ )
375 t0 = f2[4]*f2[8] - f2[5]*f2[7];
376 t1 = f2[3]*f2[8] - f2[5]*f2[6];
377 t2 = f2[3]*f2[7] - f2[4]*f2[6];
379 c[3] = f2[0]*t0 - f2[1]*t1 + f2[2]*t2;
381 c[2] = f1[0]*t0 - f1[1]*t1 + f1[2]*t2 -
382 f1[3]*(f2[1]*f2[8] - f2[2]*f2[7]) +
383 f1[4]*(f2[0]*f2[8] - f2[2]*f2[6]) -
384 f1[5]*(f2[0]*f2[7] - f2[1]*f2[6]) +
385 f1[6]*(f2[1]*f2[5] - f2[2]*f2[4]) -
386 f1[7]*(f2[0]*f2[5] - f2[2]*f2[3]) +
387 f1[8]*(f2[0]*f2[4] - f2[1]*f2[3]);
389 t0 = f1[4]*f1[8] - f1[5]*f1[7];
390 t1 = f1[3]*f1[8] - f1[5]*f1[6];
391 t2 = f1[3]*f1[7] - f1[4]*f1[6];
393 c[1] = f2[0]*t0 - f2[1]*t1 + f2[2]*t2 -
394 f2[3]*(f1[1]*f1[8] - f1[2]*f1[7]) +
395 f2[4]*(f1[0]*f1[8] - f1[2]*f1[6]) -
396 f2[5]*(f1[0]*f1[7] - f1[1]*f1[6]) +
397 f2[6]*(f1[1]*f1[5] - f1[2]*f1[4]) -
398 f2[7]*(f1[0]*f1[5] - f1[2]*f1[3]) +
399 f2[8]*(f1[0]*f1[4] - f1[1]*f1[3]);
401 c[0] = f1[0]*t0 - f1[1]*t1 + f1[2]*t2;
403 // solve the cubic equation; there can be 1 to 3 roots ...
404 n = cvSolveCubic( &coeffs, &roots );
409 for( k = 0; k < n; k++, fmatrix += 9 )
411 // for each root form the fundamental matrix
412 double lambda = r[k], mu = 1.;
413 double s = f1[8]*r[k] + f2[8];
415 // normalize each matrix, so that F(3,3) (~fmatrix[8]) == 1
416 if( fabs(s) > DBL_EPSILON )
425 for( i = 0; i < 8; i++ )
426 fmatrix[i] = f1[i]*lambda + f2[i]*mu;
433 int CvFMEstimator::run8Point( const CvMat* _m1, const CvMat* _m2, CvMat* _fmatrix )
435 double a[9*9], w[9], v[9*9];
436 CvMat W = cvMat( 1, 9, CV_64F, w );
437 CvMat V = cvMat( 9, 9, CV_64F, v );
438 CvMat A = cvMat( 9, 9, CV_64F, a );
441 CvPoint2D64f m0c = {0,0}, m1c = {0,0};
442 double t, scale0 = 0, scale1 = 0;
444 const CvPoint2D64f* m1 = (const CvPoint2D64f*)_m1->data.ptr;
445 const CvPoint2D64f* m2 = (const CvPoint2D64f*)_m2->data.ptr;
446 double* fmatrix = _fmatrix->data.db;
447 CV_Assert( (_m1->cols == 1 || _m1->rows == 1) && CV_ARE_SIZES_EQ(_m1, _m2));
448 int i, j, k, count = _m1->cols*_m1->rows;
450 // compute centers and average distances for each of the two point sets
451 for( i = 0; i < count; i++ )
453 double x = m1[i].x, y = m1[i].y;
454 m0c.x += x; m0c.y += y;
456 x = m2[i].x, y = m2[i].y;
457 m1c.x += x; m1c.y += y;
460 // calculate the normalizing transformations for each of the point sets:
461 // after the transformation each set will have the mass center at the coordinate origin
462 // and the average distance from the origin will be ~sqrt(2).
464 m0c.x *= t; m0c.y *= t;
465 m1c.x *= t; m1c.y *= t;
467 for( i = 0; i < count; i++ )
469 double x = m1[i].x - m0c.x, y = m1[i].y - m0c.y;
470 scale0 += sqrt(x*x + y*y);
472 x = m2[i].x - m1c.x, y = m2[i].y - m1c.y;
473 scale1 += sqrt(x*x + y*y);
479 if( scale0 < FLT_EPSILON || scale1 < FLT_EPSILON )
482 scale0 = sqrt(2.)/scale0;
483 scale1 = sqrt(2.)/scale1;
487 // form a linear system Ax=0: for each selected pair of points m1 & m2,
488 // the row of A(=a) represents the coefficients of equation: (m2, 1)'*F*(m1, 1) = 0
489 // to save computation time, we compute (At*A) instead of A and then solve (At*A)x=0.
490 for( i = 0; i < count; i++ )
492 double x0 = (m1[i].x - m0c.x)*scale0;
493 double y0 = (m1[i].y - m0c.y)*scale0;
494 double x1 = (m2[i].x - m1c.x)*scale1;
495 double y1 = (m2[i].y - m1c.y)*scale1;
496 double r[9] = { x1*x0, x1*y0, x1, y1*x0, y1*y0, y1, x0, y0, 1 };
497 for( j = 0; j < 9; j++ )
498 for( k = 0; k < 9; k++ )
499 a[j*9+k] += r[j]*r[k];
502 cvEigenVV(&A, &V, &W);
504 for( i = 0; i < 9; i++ )
506 if( fabs(w[i]) < DBL_EPSILON )
513 F0 = cvMat( 3, 3, CV_64F, v + 9*8 ); // take the last column of v as a solution of Af = 0
515 // make F0 singular (of rank 2) by decomposing it with SVD,
516 // zeroing the last diagonal element of W and then composing the matrices back.
518 // use v as a temporary storage for different 3x3 matrices
525 cvSVD( &F0, &W, &U, &V, CV_SVD_MODIFY_A + CV_SVD_U_T + CV_SVD_V_T );
528 // F0 <- U*diag([W(1), W(2), 0])*V'
529 cvGEMM( &U, &W, 1., 0, 0., &TF, CV_GEMM_A_T );
530 cvGEMM( &TF, &V, 1., 0, 0., &F0, 0/*CV_GEMM_B_T*/ );
532 // apply the transformation that is inverse
533 // to what we used to normalize the point coordinates
535 double tt0[] = { scale0, 0, -scale0*m0c.x, 0, scale0, -scale0*m0c.y, 0, 0, 1 };
536 double tt1[] = { scale1, 0, -scale1*m1c.x, 0, scale1, -scale1*m1c.y, 0, 0, 1 };
543 cvGEMM( &T1, &F0, 1., 0, 0., &TF, CV_GEMM_A_T );
544 F0.data.db = fmatrix;
545 cvGEMM( &TF, &T0, 1., 0, 0., &F0, 0 );
548 if( fabs(F0.data.db[8]) > FLT_EPSILON )
549 cvScale( &F0, &F0, 1./F0.data.db[8] );
556 void CvFMEstimator::computeReprojError( const CvMat* _m1, const CvMat* _m2,
557 const CvMat* model, CvMat* _err )
559 int i, count = _m1->rows*_m1->cols;
560 const CvPoint2D64f* m1 = (const CvPoint2D64f*)_m1->data.ptr;
561 const CvPoint2D64f* m2 = (const CvPoint2D64f*)_m2->data.ptr;
562 const double* F = model->data.db;
563 float* err = _err->data.fl;
565 for( i = 0; i < count; i++ )
567 double a, b, c, d1, d2, s1, s2;
569 a = F[0]*m1[i].x + F[1]*m1[i].y + F[2];
570 b = F[3]*m1[i].x + F[4]*m1[i].y + F[5];
571 c = F[6]*m1[i].x + F[7]*m1[i].y + F[8];
574 d2 = m2[i].x*a + m2[i].y*b + c;
576 a = F[0]*m2[i].x + F[3]*m2[i].y + F[6];
577 b = F[1]*m2[i].x + F[4]*m2[i].y + F[7];
578 c = F[2]*m2[i].x + F[5]*m2[i].y + F[8];
581 d1 = m1[i].x*a + m1[i].y*b + c;
583 err[i] = (float)std::max(d1*d1*s1, d2*d2*s2);
588 CV_IMPL int cvFindFundamentalMat( const CvMat* points1, const CvMat* points2,
589 CvMat* fmatrix, int method,
590 double param1, double param2, CvMat* mask )
593 Ptr<CvMat> m1, m2, tempMask;
596 CvMat _F3x3 = cvMat( 3, 3, CV_64FC1, F ), _F9x3 = cvMat( 9, 3, CV_64FC1, F );
599 CV_Assert( CV_IS_MAT(points1) && CV_IS_MAT(points2) && CV_ARE_SIZES_EQ(points1, points2) );
600 CV_Assert( CV_IS_MAT(fmatrix) && fmatrix->cols == 3 &&
601 (fmatrix->rows == 3 || (fmatrix->rows == 9 && method == CV_FM_7POINT)) );
603 count = MAX(points1->cols, points1->rows);
607 m1 = cvCreateMat( 1, count, CV_64FC2 );
608 cvConvertPointsHomogeneous( points1, m1 );
610 m2 = cvCreateMat( 1, count, CV_64FC2 );
611 cvConvertPointsHomogeneous( points2, m2 );
615 CV_Assert( CV_IS_MASK_ARR(mask) && CV_IS_MAT_CONT(mask->type) &&
616 (mask->rows == 1 || mask->cols == 1) &&
617 mask->rows*mask->cols == count );
619 if( mask || count >= 8 )
620 tempMask = cvCreateMat( 1, count, CV_8U );
621 if( !tempMask.empty() )
622 cvSet( tempMask, cvScalarAll(1.) );
624 CvFMEstimator estimator(7);
626 result = estimator.run7Point(m1, m2, &_F9x3);
627 else if( method == CV_FM_8POINT )
628 result = estimator.run8Point(m1, m2, &_F3x3);
633 if( param2 < DBL_EPSILON || param2 > 1 - DBL_EPSILON )
636 if( (method & ~3) == CV_RANSAC && count >= 15 )
637 result = estimator.runRANSAC(m1, m2, &_F3x3, tempMask, param1, param2 );
639 result = estimator.runLMeDS(m1, m2, &_F3x3, tempMask, param2 );
642 /*icvCompressPoints( (CvPoint2D64f*)m1->data.ptr, tempMask->data.ptr, 1, count );
643 count = icvCompressPoints( (CvPoint2D64f*)m2->data.ptr, tempMask->data.ptr, 1, count );
644 assert( count >= 8 );
645 m1->cols = m2->cols = count;
646 estimator.run8Point(m1, m2, &_F3x3);*/
650 cvConvert( fmatrix->rows == 3 ? &_F3x3 : &_F9x3, fmatrix );
652 if( mask && tempMask )
654 if( CV_ARE_SIZES_EQ(mask, tempMask) )
655 cvCopy( tempMask, mask );
657 cvTranspose( tempMask, mask );
664 CV_IMPL void cvComputeCorrespondEpilines( const CvMat* points, int pointImageID,
665 const CvMat* fmatrix, CvMat* lines )
667 int abc_stride, abc_plane_stride, abc_elem_size;
668 int plane_stride, stride, elem_size;
669 int i, dims, count, depth, cn, abc_dims, abc_count, abc_depth, abc_cn;
671 const uchar *xp, *yp, *zp;
673 CvMat F = cvMat( 3, 3, CV_64F, f );
675 if( !CV_IS_MAT(points) )
676 CV_Error( !points ? CV_StsNullPtr : CV_StsBadArg, "points parameter is not a valid matrix" );
678 depth = CV_MAT_DEPTH(points->type);
679 cn = CV_MAT_CN(points->type);
680 if( (depth != CV_32F && depth != CV_64F) || (cn != 1 && cn != 2 && cn != 3) )
681 CV_Error( CV_StsUnsupportedFormat, "The format of point matrix is unsupported" );
686 CV_Assert( points->rows == 1 || points->cols == 1 );
687 count = points->rows * points->cols;
689 else if( points->rows > points->cols )
691 dims = cn*points->cols;
692 count = points->rows;
696 if( (points->rows > 1 && cn > 1) || (points->rows == 1 && cn == 1) )
697 CV_Error( CV_StsBadSize, "The point matrix does not have a proper layout (2xn, 3xn, nx2 or nx3)" );
699 count = points->cols;
702 if( dims != 2 && dims != 3 )
703 CV_Error( CV_StsOutOfRange, "The dimensionality of points must be 2 or 3" );
705 if( !CV_IS_MAT(fmatrix) )
706 CV_Error( !fmatrix ? CV_StsNullPtr : CV_StsBadArg, "fmatrix is not a valid matrix" );
708 if( CV_MAT_TYPE(fmatrix->type) != CV_32FC1 && CV_MAT_TYPE(fmatrix->type) != CV_64FC1 )
709 CV_Error( CV_StsUnsupportedFormat, "fundamental matrix must have 32fC1 or 64fC1 type" );
711 if( fmatrix->cols != 3 || fmatrix->rows != 3 )
712 CV_Error( CV_StsBadSize, "fundamental matrix must be 3x3" );
714 if( !CV_IS_MAT(lines) )
715 CV_Error( !lines ? CV_StsNullPtr : CV_StsBadArg, "lines parameter is not a valid matrix" );
717 abc_depth = CV_MAT_DEPTH(lines->type);
718 abc_cn = CV_MAT_CN(lines->type);
719 if( (abc_depth != CV_32F && abc_depth != CV_64F) || (abc_cn != 1 && abc_cn != 3) )
720 CV_Error( CV_StsUnsupportedFormat, "The format of the matrix of lines is unsupported" );
725 CV_Assert( lines->rows == 1 || lines->cols == 1 );
726 abc_count = lines->rows * lines->cols;
728 else if( lines->rows > lines->cols )
730 abc_dims = abc_cn*lines->cols;
731 abc_count = lines->rows;
735 if( (lines->rows > 1 && abc_cn > 1) || (lines->rows == 1 && abc_cn == 1) )
736 CV_Error( CV_StsBadSize, "The lines matrix does not have a proper layout (3xn or nx3)" );
737 abc_dims = lines->rows;
738 abc_count = lines->cols;
742 CV_Error( CV_StsOutOfRange, "The lines matrix does not have a proper layout (3xn or nx3)" );
744 if( abc_count != count )
745 CV_Error( CV_StsUnmatchedSizes, "The numbers of points and lines are different" );
747 elem_size = CV_ELEM_SIZE(depth);
748 abc_elem_size = CV_ELEM_SIZE(abc_depth);
750 if( cn == 1 && points->rows == dims )
752 plane_stride = points->step;
757 plane_stride = elem_size;
758 stride = points->rows == 1 ? dims*elem_size : points->step;
761 if( abc_cn == 1 && lines->rows == 3 )
763 abc_plane_stride = lines->step;
764 abc_stride = abc_elem_size;
768 abc_plane_stride = abc_elem_size;
769 abc_stride = lines->rows == 1 ? 3*abc_elem_size : lines->step;
772 cvConvert( fmatrix, &F );
773 if( pointImageID == 2 )
774 cvTranspose( &F, &F );
776 xp = points->data.ptr;
777 yp = xp + plane_stride;
778 zp = dims == 3 ? yp + plane_stride : 0;
780 ap = lines->data.ptr;
781 bp = ap + abc_plane_stride;
782 cp = bp + abc_plane_stride;
784 for( i = 0; i < count; i++ )
789 if( depth == CV_32F )
791 x = *(float*)xp; y = *(float*)yp;
793 z = *(float*)zp, zp += stride;
797 x = *(double*)xp; y = *(double*)yp;
799 z = *(double*)zp, zp += stride;
802 xp += stride; yp += stride;
804 a = f[0]*x + f[1]*y + f[2]*z;
805 b = f[3]*x + f[4]*y + f[5]*z;
806 c = f[6]*x + f[7]*y + f[8]*z;
808 nu = nu ? 1./sqrt(nu) : 1.;
809 a *= nu; b *= nu; c *= nu;
811 if( abc_depth == CV_32F )
813 *(float*)ap = (float)a;
814 *(float*)bp = (float)b;
815 *(float*)cp = (float)c;
831 CV_IMPL void cvConvertPointsHomogeneous( const CvMat* src, CvMat* dst )
833 Ptr<CvMat> temp, denom;
835 int i, s_count, s_dims, d_count, d_dims;
836 CvMat _src, _dst, _ones;
839 if( !CV_IS_MAT(src) )
840 CV_Error( !src ? CV_StsNullPtr : CV_StsBadArg,
841 "The input parameter is not a valid matrix" );
843 if( !CV_IS_MAT(dst) )
844 CV_Error( !dst ? CV_StsNullPtr : CV_StsBadArg,
845 "The output parameter is not a valid matrix" );
847 if( src == dst || src->data.ptr == dst->data.ptr )
849 if( src != dst && (!CV_ARE_TYPES_EQ(src, dst) || !CV_ARE_SIZES_EQ(src,dst)) )
850 CV_Error( CV_StsBadArg, "Invalid inplace operation" );
854 if( src->rows > src->cols )
856 if( !((src->cols > 1) ^ (CV_MAT_CN(src->type) > 1)) )
857 CV_Error( CV_StsBadSize, "Either the number of channels or columns or rows must be =1" );
859 s_dims = CV_MAT_CN(src->type)*src->cols;
864 if( !((src->rows > 1) ^ (CV_MAT_CN(src->type) > 1)) )
865 CV_Error( CV_StsBadSize, "Either the number of channels or columns or rows must be =1" );
867 s_dims = CV_MAT_CN(src->type)*src->rows;
871 if( src->rows == 1 || src->cols == 1 )
872 src = cvReshape( src, &_src, 1, s_count );
874 if( dst->rows > dst->cols )
876 if( !((dst->cols > 1) ^ (CV_MAT_CN(dst->type) > 1)) )
877 CV_Error( CV_StsBadSize,
878 "Either the number of channels or columns or rows in the input matrix must be =1" );
880 d_dims = CV_MAT_CN(dst->type)*dst->cols;
885 if( !((dst->rows > 1) ^ (CV_MAT_CN(dst->type) > 1)) )
886 CV_Error( CV_StsBadSize,
887 "Either the number of channels or columns or rows in the output matrix must be =1" );
889 d_dims = CV_MAT_CN(dst->type)*dst->rows;
893 if( dst->rows == 1 || dst->cols == 1 )
894 dst = cvReshape( dst, &_dst, 1, d_count );
896 if( s_count != d_count )
897 CV_Error( CV_StsUnmatchedSizes, "Both matrices must have the same number of points" );
899 if( CV_MAT_DEPTH(src->type) < CV_32F || CV_MAT_DEPTH(dst->type) < CV_32F )
900 CV_Error( CV_StsUnsupportedFormat,
901 "Both matrices must be floating-point (single or double precision)" );
903 if( s_dims < 2 || s_dims > 4 || d_dims < 2 || d_dims > 4 )
904 CV_Error( CV_StsOutOfRange,
905 "Both input and output point dimensionality must be 2, 3 or 4" );
907 if( s_dims < d_dims - 1 || s_dims > d_dims + 1 )
908 CV_Error( CV_StsUnmatchedSizes,
909 "The dimensionalities of input and output point sets differ too much" );
911 if( s_dims == d_dims - 1 )
913 if( d_count == dst->rows )
915 ones = cvGetSubRect( dst, &_ones, cvRect( s_dims, 0, 1, d_count ));
916 dst = cvGetSubRect( dst, &_dst, cvRect( 0, 0, s_dims, d_count ));
920 ones = cvGetSubRect( dst, &_ones, cvRect( 0, s_dims, d_count, 1 ));
921 dst = cvGetSubRect( dst, &_dst, cvRect( 0, 0, d_count, s_dims ));
925 if( s_dims <= d_dims )
927 if( src->rows == dst->rows && src->cols == dst->cols )
929 if( CV_ARE_TYPES_EQ( src, dst ) )
932 cvConvert( src, dst );
936 if( !CV_ARE_TYPES_EQ( src, dst ))
938 temp = cvCreateMat( src->rows, src->cols, dst->type );
939 cvConvert( src, temp );
942 cvTranspose( src, dst );
946 cvSet( ones, cvRealScalar(1.) );
950 int s_plane_stride, s_stride, d_plane_stride, d_stride, elem_size;
952 if( !CV_ARE_TYPES_EQ( src, dst ))
954 temp = cvCreateMat( src->rows, src->cols, dst->type );
955 cvConvert( src, temp );
959 elem_size = CV_ELEM_SIZE(src->type);
961 if( s_count == src->cols )
962 s_plane_stride = src->step / elem_size, s_stride = 1;
964 s_stride = src->step / elem_size, s_plane_stride = 1;
966 if( d_count == dst->cols )
967 d_plane_stride = dst->step / elem_size, d_stride = 1;
969 d_stride = dst->step / elem_size, d_plane_stride = 1;
971 denom = cvCreateMat( 1, d_count, dst->type );
973 if( CV_MAT_DEPTH(dst->type) == CV_32F )
975 const float* xs = src->data.fl;
976 const float* ys = xs + s_plane_stride;
978 const float* ws = xs + (s_dims - 1)*s_plane_stride;
980 float* iw = denom->data.fl;
982 float* xd = dst->data.fl;
983 float* yd = xd + d_plane_stride;
988 zs = ys + s_plane_stride;
989 zd = yd + d_plane_stride;
992 for( i = 0; i < d_count; i++, ws += s_stride )
995 iw[i] = fabs((double)t) > FLT_EPSILON ? t : 1.f;
998 cvDiv( 0, denom, denom );
1001 for( i = 0; i < d_count; i++ )
1004 float x = *xs * w, y = *ys * w, z = *zs * w;
1005 xs += s_stride; ys += s_stride; zs += s_stride;
1006 *xd = x; *yd = y; *zd = z;
1007 xd += d_stride; yd += d_stride; zd += d_stride;
1010 for( i = 0; i < d_count; i++ )
1013 float x = *xs * w, y = *ys * w;
1014 xs += s_stride; ys += s_stride;
1016 xd += d_stride; yd += d_stride;
1021 const double* xs = src->data.db;
1022 const double* ys = xs + s_plane_stride;
1023 const double* zs = 0;
1024 const double* ws = xs + (s_dims - 1)*s_plane_stride;
1026 double* iw = denom->data.db;
1028 double* xd = dst->data.db;
1029 double* yd = xd + d_plane_stride;
1034 zs = ys + s_plane_stride;
1035 zd = yd + d_plane_stride;
1038 for( i = 0; i < d_count; i++, ws += s_stride )
1041 iw[i] = fabs(t) > DBL_EPSILON ? t : 1.;
1044 cvDiv( 0, denom, denom );
1047 for( i = 0; i < d_count; i++ )
1050 double x = *xs * w, y = *ys * w, z = *zs * w;
1051 xs += s_stride; ys += s_stride; zs += s_stride;
1052 *xd = x; *yd = y; *zd = z;
1053 xd += d_stride; yd += d_stride; zd += d_stride;
1056 for( i = 0; i < d_count; i++ )
1059 double x = *xs * w, y = *ys * w;
1060 xs += s_stride; ys += s_stride;
1062 xd += d_stride; yd += d_stride;
1068 cv::Mat cv::findHomography( InputArray _points1, InputArray _points2,
1069 int method, double ransacReprojThreshold, OutputArray _mask )
1071 Mat points1 = _points1.getMat(), points2 = _points2.getMat();
1072 int npoints = points1.checkVector(2);
1073 CV_Assert( npoints >= 0 && points2.checkVector(2) == npoints &&
1074 points1.type() == points2.type());
1076 Mat H(3, 3, CV_64F);
1077 CvMat _pt1 = points1, _pt2 = points2;
1078 CvMat matH = H, c_mask, *p_mask = 0;
1079 if( _mask.needed() )
1081 _mask.create(npoints, 1, CV_8U, -1, true);
1082 p_mask = &(c_mask = _mask.getMat());
1084 bool ok = cvFindHomography( &_pt1, &_pt2, &matH, method, ransacReprojThreshold, p_mask ) > 0;
1090 cv::Mat cv::findHomography( InputArray _points1, InputArray _points2,
1091 OutputArray _mask, int method, double ransacReprojThreshold )
1093 return cv::findHomography(_points1, _points2, method, ransacReprojThreshold, _mask);
1096 cv::Mat cv::findFundamentalMat( InputArray _points1, InputArray _points2,
1097 int method, double param1, double param2,
1100 Mat points1 = _points1.getMat(), points2 = _points2.getMat();
1101 int npoints = points1.checkVector(2);
1102 CV_Assert( npoints >= 0 && points2.checkVector(2) == npoints &&
1103 points1.type() == points2.type());
1105 Mat F(method == CV_FM_7POINT ? 9 : 3, 3, CV_64F);
1106 CvMat _pt1 = points1, _pt2 = points2;
1107 CvMat matF = F, c_mask, *p_mask = 0;
1108 if( _mask.needed() )
1110 _mask.create(npoints, 1, CV_8U, -1, true);
1111 p_mask = &(c_mask = _mask.getMat());
1113 int n = cvFindFundamentalMat( &_pt1, &_pt2, &matF, method, param1, param2, p_mask );
1117 F = F.rowRange(0, 3);
1121 cv::Mat cv::findFundamentalMat( InputArray _points1, InputArray _points2,
1122 OutputArray _mask, int method, double param1, double param2 )
1124 return cv::findFundamentalMat(_points1, _points2, method, param1, param2, _mask);
1128 void cv::computeCorrespondEpilines( InputArray _points, int whichImage,
1129 InputArray _Fmat, OutputArray _lines )
1131 Mat points = _points.getMat(), F = _Fmat.getMat();
1132 int npoints = points.checkVector(2);
1134 npoints = points.checkVector(3);
1135 CV_Assert( npoints >= 0 && (points.depth() == CV_32F || points.depth() == CV_32S));
1137 _lines.create(npoints, 1, CV_32FC3, -1, true);
1138 CvMat c_points = points, c_lines = _lines.getMat(), c_F = F;
1139 cvComputeCorrespondEpilines(&c_points, whichImage, &c_F, &c_lines);
1142 void cv::convertPointsFromHomogeneous( InputArray _src, OutputArray _dst )
1144 Mat src = _src.getMat();
1145 int npoints = src.checkVector(3), cn = 3;
1148 npoints = src.checkVector(4);
1152 CV_Assert( npoints >= 0 && (src.depth() == CV_32F || src.depth() == CV_32S));
1154 _dst.create(npoints, 1, CV_MAKETYPE(CV_32F, cn-1));
1155 CvMat c_src = src, c_dst = _dst.getMat();
1156 cvConvertPointsHomogeneous(&c_src, &c_dst);
1159 void cv::convertPointsToHomogeneous( InputArray _src, OutputArray _dst )
1161 Mat src = _src.getMat();
1162 int npoints = src.checkVector(2), cn = 2;
1165 npoints = src.checkVector(3);
1169 CV_Assert( npoints >= 0 && (src.depth() == CV_32F || src.depth() == CV_32S));
1171 _dst.create(npoints, 1, CV_MAKETYPE(CV_32F, cn+1));
1172 CvMat c_src = src, c_dst = _dst.getMat();
1173 cvConvertPointsHomogeneous(&c_src, &c_dst);
1176 void cv::convertPointsHomogeneous( InputArray _src, OutputArray _dst )
1178 int stype = _src.type(), dtype = _dst.type();
1179 CV_Assert( _dst.fixedType() );
1181 if( CV_MAT_CN(stype) > CV_MAT_CN(dtype) )
1182 convertPointsFromHomogeneous(_src, _dst);
1184 convertPointsToHomogeneous(_src, _dst);