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 //                        Intel License Agreement
11 //
12 // Copyright (C) 2000, Intel Corporation, all rights reserved.
13 // Third party copyrights are property of their respective owners.
14 //
15 // Redistribution and use in source and binary forms, with or without modification,
16 // are permitted provided that the following conditions are met:
17 //
18 //   * Redistribution's of source code must retain the above copyright notice,
19 //     this list of conditions and the following disclaimer.
20 //
21 //   * Redistribution's in binary form must reproduce the above copyright notice,
22 //     this list of conditions and the following disclaimer in the documentation
23 //     and/or other materials provided with the distribution.
24 //
25 //   * The name of Intel Corporation may not be used to endorse or promote products
26 //     derived from this software without specific prior written permission.
27 //
28 // This software is provided by the copyright holders and contributors "as is" and
29 // any express or implied warranties, including, but not limited to, the implied
30 // warranties of merchantability and fitness for a particular purpose are disclaimed.
31 // In no event shall the Intel Corporation or contributors be liable for any direct,
32 // indirect, incidental, special, exemplary, or consequential damages
33 // (including, but not limited to, procurement of substitute goods or services;
34 // loss of use, data, or profits; or business interruption) however caused
35 // and on any theory of liability, whether in contract, strict liability,
36 // or tort (including negligence or otherwise) arising in any way out of
37 // the use of this software, even if advised of the possibility of such damage.
38 //
39 //M*/
40 
41 #include "old_ml_precomp.hpp"
42 
43 
CvStatModel()44 CvStatModel::CvStatModel()
45 {
46     default_model_name = "my_stat_model";
47 }
48 
49 
~CvStatModel()50 CvStatModel::~CvStatModel()
51 {
52     clear();
53 }
54 
55 
clear()56 void CvStatModel::clear()
57 {
58 }
59 
60 
save(const char * filename,const char * name) const61 void CvStatModel::save( const char* filename, const char* name ) const
62 {
63     CvFileStorage* fs = 0;
64 
65     CV_FUNCNAME( "CvStatModel::save" );
66 
67     __BEGIN__;
68 
69     CV_CALL( fs = cvOpenFileStorage( filename, 0, CV_STORAGE_WRITE ));
70     if( !fs )
71         CV_ERROR( CV_StsError, "Could not open the file storage. Check the path and permissions" );
72 
73     write( fs, name ? name : default_model_name );
74 
75     __END__;
76 
77     cvReleaseFileStorage( &fs );
78 }
79 
80 
load(const char * filename,const char * name)81 void CvStatModel::load( const char* filename, const char* name )
82 {
83     CvFileStorage* fs = 0;
84 
85     CV_FUNCNAME( "CvAlgorithm::load" );
86 
87     __BEGIN__;
88 
89     CvFileNode* model_node = 0;
90 
91     CV_CALL( fs = cvOpenFileStorage( filename, 0, CV_STORAGE_READ ));
92     if( !fs )
93         EXIT;
94 
95     if( name )
96         model_node = cvGetFileNodeByName( fs, 0, name );
97     else
98     {
99         CvFileNode* root = cvGetRootFileNode( fs );
100         if( root->data.seq->total > 0 )
101             model_node = (CvFileNode*)cvGetSeqElem( root->data.seq, 0 );
102     }
103 
104     read( fs, model_node );
105 
106     __END__;
107 
108     cvReleaseFileStorage( &fs );
109 }
110 
111 
write(CvFileStorage *,const char *) const112 void CvStatModel::write( CvFileStorage*, const char* ) const
113 {
114     OPENCV_ERROR( CV_StsNotImplemented, "CvStatModel::write", "" );
115 }
116 
117 
read(CvFileStorage *,CvFileNode *)118 void CvStatModel::read( CvFileStorage*, CvFileNode* )
119 {
120     OPENCV_ERROR( CV_StsNotImplemented, "CvStatModel::read", "" );
121 }
122 
123 
124 /* Calculates upper triangular matrix S, where A is a symmetrical matrix A=S'*S */
cvChol(CvMat * A,CvMat * S)125 static void cvChol( CvMat* A, CvMat* S )
126 {
127     int dim = A->rows;
128 
129     int i, j, k;
130     float sum;
131 
132     for( i = 0; i < dim; i++ )
133     {
134         for( j = 0; j < i; j++ )
135             CV_MAT_ELEM(*S, float, i, j) = 0;
136 
137         sum = 0;
138         for( k = 0; k < i; k++ )
139             sum += CV_MAT_ELEM(*S, float, k, i) * CV_MAT_ELEM(*S, float, k, i);
140 
141         CV_MAT_ELEM(*S, float, i, i) = (float)sqrt(CV_MAT_ELEM(*A, float, i, i) - sum);
142 
143         for( j = i + 1; j < dim; j++ )
144         {
145             sum = 0;
146             for( k = 0; k < i; k++ )
147                 sum += CV_MAT_ELEM(*S, float, k, i) * CV_MAT_ELEM(*S, float, k, j);
148 
149             CV_MAT_ELEM(*S, float, i, j) =
150                 (CV_MAT_ELEM(*A, float, i, j) - sum) / CV_MAT_ELEM(*S, float, i, i);
151 
152         }
153     }
154 }
155 
156 /* Generates <sample> from multivariate normal distribution, where <mean> - is an
157    average row vector, <cov> - symmetric covariation matrix */
cvRandMVNormal(CvMat * mean,CvMat * cov,CvMat * sample,CvRNG * rng)158 CV_IMPL void cvRandMVNormal( CvMat* mean, CvMat* cov, CvMat* sample, CvRNG* rng )
159 {
160     int dim = sample->cols;
161     int amount = sample->rows;
162 
163     CvRNG state = rng ? *rng : cvRNG( cvGetTickCount() );
164     cvRandArr(&state, sample, CV_RAND_NORMAL, cvScalarAll(0), cvScalarAll(1) );
165 
166     CvMat* utmat = cvCreateMat(dim, dim, sample->type);
167     CvMat* vect = cvCreateMatHeader(1, dim, sample->type);
168 
169     cvChol(cov, utmat);
170 
171     int i;
172     for( i = 0; i < amount; i++ )
173     {
174         cvGetRow(sample, vect, i);
175         cvMatMulAdd(vect, utmat, mean, vect);
176     }
177 
178     cvReleaseMat(&vect);
179     cvReleaseMat(&utmat);
180 }
181 
182 
183 /* Generates <sample> of <amount> points from a discrete variate xi,
184    where Pr{xi = k} == probs[k], 0 < k < len - 1. */
cvRandSeries(float probs[],int len,int sample[],int amount)185 static void cvRandSeries( float probs[], int len, int sample[], int amount )
186 {
187     CvMat* univals = cvCreateMat(1, amount, CV_32FC1);
188     float* knots = (float*)cvAlloc( len * sizeof(float) );
189 
190     int i, j;
191 
192     CvRNG state = cvRNG(-1);
193     cvRandArr(&state, univals, CV_RAND_UNI, cvScalarAll(0), cvScalarAll(1) );
194 
195     knots[0] = probs[0];
196     for( i = 1; i < len; i++ )
197         knots[i] = knots[i - 1] + probs[i];
198 
199     for( i = 0; i < amount; i++ )
200         for( j = 0; j < len; j++ )
201         {
202             if ( CV_MAT_ELEM(*univals, float, 0, i) <= knots[j] )
203             {
204                 sample[i] = j;
205                 break;
206             }
207         }
208 
209     cvFree(&knots);
210 }
211 
212 /* Generates <sample> from gaussian mixture distribution */
cvRandGaussMixture(CvMat * means[],CvMat * covs[],float weights[],int clsnum,CvMat * sample,CvMat * sampClasses)213 CV_IMPL void cvRandGaussMixture( CvMat* means[],
214                                  CvMat* covs[],
215                                  float weights[],
216                                  int clsnum,
217                                  CvMat* sample,
218                                  CvMat* sampClasses )
219 {
220     int dim = sample->cols;
221     int amount = sample->rows;
222 
223     int i, clss;
224 
225     int* sample_clsnum = (int*)cvAlloc( amount * sizeof(int) );
226     CvMat** utmats = (CvMat**)cvAlloc( clsnum * sizeof(CvMat*) );
227     CvMat* vect = cvCreateMatHeader(1, dim, CV_32FC1);
228 
229     CvMat* classes;
230     if( sampClasses )
231         classes = sampClasses;
232     else
233         classes = cvCreateMat(1, amount, CV_32FC1);
234 
235     CvRNG state = cvRNG(-1);
236     cvRandArr(&state, sample, CV_RAND_NORMAL, cvScalarAll(0), cvScalarAll(1));
237 
238     cvRandSeries(weights, clsnum, sample_clsnum, amount);
239 
240     for( i = 0; i < clsnum; i++ )
241     {
242         utmats[i] = cvCreateMat(dim, dim, CV_32FC1);
243         cvChol(covs[i], utmats[i]);
244     }
245 
246     for( i = 0; i < amount; i++ )
247     {
248         CV_MAT_ELEM(*classes, float, 0, i) = (float)sample_clsnum[i];
249         cvGetRow(sample, vect, i);
250         clss = sample_clsnum[i];
251         cvMatMulAdd(vect, utmats[clss], means[clss], vect);
252     }
253 
254     if( !sampClasses )
255         cvReleaseMat(&classes);
256     for( i = 0; i < clsnum; i++ )
257         cvReleaseMat(&utmats[i]);
258     cvFree(&utmats);
259     cvFree(&sample_clsnum);
260     cvReleaseMat(&vect);
261 }
262 
263 
icvGenerateRandomClusterCenters(int seed,const CvMat * data,int num_of_clusters,CvMat * _centers)264 CvMat* icvGenerateRandomClusterCenters ( int seed, const CvMat* data,
265                                          int num_of_clusters, CvMat* _centers )
266 {
267     CvMat* centers = _centers;
268 
269     CV_FUNCNAME("icvGenerateRandomClusterCenters");
270     __BEGIN__;
271 
272     CvRNG rng;
273     CvMat data_comp, centers_comp;
274     CvPoint minLoc, maxLoc; // Not used, just for function "cvMinMaxLoc"
275     double minVal, maxVal;
276     int i;
277     int dim = data ? data->cols : 0;
278 
279     if( ICV_IS_MAT_OF_TYPE(data, CV_32FC1) )
280     {
281         if( _centers && !ICV_IS_MAT_OF_TYPE (_centers, CV_32FC1) )
282         {
283             CV_ERROR(CV_StsBadArg,"");
284         }
285         else if( !_centers )
286             CV_CALL(centers = cvCreateMat (num_of_clusters, dim, CV_32FC1));
287     }
288     else if( ICV_IS_MAT_OF_TYPE(data, CV_64FC1) )
289     {
290         if( _centers && !ICV_IS_MAT_OF_TYPE (_centers, CV_64FC1) )
291         {
292             CV_ERROR(CV_StsBadArg,"");
293         }
294         else if( !_centers )
295             CV_CALL(centers = cvCreateMat (num_of_clusters, dim, CV_64FC1));
296     }
297     else
298         CV_ERROR (CV_StsBadArg,"");
299 
300     if( num_of_clusters < 1 )
301         CV_ERROR (CV_StsBadArg,"");
302 
303     rng = cvRNG(seed);
304     for (i = 0; i < dim; i++)
305     {
306         CV_CALL(cvGetCol (data, &data_comp, i));
307         CV_CALL(cvMinMaxLoc (&data_comp, &minVal, &maxVal, &minLoc, &maxLoc));
308         CV_CALL(cvGetCol (centers, &centers_comp, i));
309         CV_CALL(cvRandArr (&rng, &centers_comp, CV_RAND_UNI, cvScalarAll(minVal), cvScalarAll(maxVal)));
310     }
311 
312     __END__;
313 
314     if( (cvGetErrStatus () < 0) || (centers != _centers) )
315         cvReleaseMat (&centers);
316 
317     return _centers ? _centers : centers;
318 } // end of icvGenerateRandomClusterCenters
319 
320 // By S. Dilman - begin -
321 
322 #define ICV_RAND_MAX    4294967296 // == 2^32
323 
324 // static void cvRandRoundUni (CvMat* center,
325 //                              float radius_small,
326 //                              float radius_large,
327 //                              CvMat* desired_matrix,
328 //                              CvRNG* rng_state_ptr)
329 // {
330 //     float rad, norm, coefficient;
331 //     int dim, size, i, j;
332 //     CvMat *cov, sample;
333 //     CvRNG rng_local;
334 
335 //     CV_FUNCNAME("cvRandRoundUni");
336 //     __BEGIN__
337 
338 //     rng_local = *rng_state_ptr;
339 
340 //     CV_ASSERT ((radius_small >= 0) &&
341 //                (radius_large > 0) &&
342 //                (radius_small <= radius_large));
343 //     CV_ASSERT (center && desired_matrix && rng_state_ptr);
344 //     CV_ASSERT (center->rows == 1);
345 //     CV_ASSERT (center->cols == desired_matrix->cols);
346 
347 //     dim = desired_matrix->cols;
348 //     size = desired_matrix->rows;
349 //     cov = cvCreateMat (dim, dim, CV_32FC1);
350 //     cvSetIdentity (cov);
351 //     cvRandMVNormal (center, cov, desired_matrix, &rng_local);
352 
353 //     for (i = 0; i < size; i++)
354 //     {
355 //         rad = (float)(cvRandReal(&rng_local)*(radius_large - radius_small) + radius_small);
356 //         cvGetRow (desired_matrix, &sample, i);
357 //         norm = (float) cvNorm (&sample, 0, CV_L2);
358 //         coefficient = rad / norm;
359 //         for (j = 0; j < dim; j++)
360 //              CV_MAT_ELEM (sample, float, 0, j) *= coefficient;
361 //     }
362 
363 //     __END__
364 
365 // }
366 
367 // By S. Dilman - end -
368 
369 static int CV_CDECL
icvCmpIntegers(const void * a,const void * b)370 icvCmpIntegers( const void* a, const void* b )
371 {
372     return *(const int*)a - *(const int*)b;
373 }
374 
375 
376 static int CV_CDECL
icvCmpIntegersPtr(const void * _a,const void * _b)377 icvCmpIntegersPtr( const void* _a, const void* _b )
378 {
379     int a = **(const int**)_a;
380     int b = **(const int**)_b;
381     return (a < b ? -1 : 0)|(a > b);
382 }
383 
384 
icvCmpSparseVecElems(const void * a,const void * b)385 static int icvCmpSparseVecElems( const void* a, const void* b )
386 {
387     return ((CvSparseVecElem32f*)a)->idx - ((CvSparseVecElem32f*)b)->idx;
388 }
389 
390 
391 CvMat*
cvPreprocessIndexArray(const CvMat * idx_arr,int data_arr_size,bool check_for_duplicates)392 cvPreprocessIndexArray( const CvMat* idx_arr, int data_arr_size, bool check_for_duplicates )
393 {
394     CvMat* idx = 0;
395 
396     CV_FUNCNAME( "cvPreprocessIndexArray" );
397 
398     __BEGIN__;
399 
400     int i, idx_total, idx_selected = 0, step, type, prev = INT_MIN, is_sorted = 1;
401     uchar* srcb = 0;
402     int* srci = 0;
403     int* dsti;
404 
405     if( !CV_IS_MAT(idx_arr) )
406         CV_ERROR( CV_StsBadArg, "Invalid index array" );
407 
408     if( idx_arr->rows != 1 && idx_arr->cols != 1 )
409         CV_ERROR( CV_StsBadSize, "the index array must be 1-dimensional" );
410 
411     idx_total = idx_arr->rows + idx_arr->cols - 1;
412     srcb = idx_arr->data.ptr;
413     srci = idx_arr->data.i;
414 
415     type = CV_MAT_TYPE(idx_arr->type);
416     step = CV_IS_MAT_CONT(idx_arr->type) ? 1 : idx_arr->step/CV_ELEM_SIZE(type);
417 
418     switch( type )
419     {
420     case CV_8UC1:
421     case CV_8SC1:
422         // idx_arr is array of 1's and 0's -
423         // i.e. it is a mask of the selected components
424         if( idx_total != data_arr_size )
425             CV_ERROR( CV_StsUnmatchedSizes,
426             "Component mask should contain as many elements as the total number of input variables" );
427 
428         for( i = 0; i < idx_total; i++ )
429             idx_selected += srcb[i*step] != 0;
430 
431         if( idx_selected == 0 )
432             CV_ERROR( CV_StsOutOfRange, "No components/input_variables is selected!" );
433 
434         break;
435     case CV_32SC1:
436         // idx_arr is array of integer indices of selected components
437         if( idx_total > data_arr_size )
438             CV_ERROR( CV_StsOutOfRange,
439             "index array may not contain more elements than the total number of input variables" );
440         idx_selected = idx_total;
441         // check if sorted already
442         for( i = 0; i < idx_total; i++ )
443         {
444             int val = srci[i*step];
445             if( val >= prev )
446             {
447                 is_sorted = 0;
448                 break;
449             }
450             prev = val;
451         }
452         break;
453     default:
454         CV_ERROR( CV_StsUnsupportedFormat, "Unsupported index array data type "
455                                            "(it should be 8uC1, 8sC1 or 32sC1)" );
456     }
457 
458     CV_CALL( idx = cvCreateMat( 1, idx_selected, CV_32SC1 ));
459     dsti = idx->data.i;
460 
461     if( type < CV_32SC1 )
462     {
463         for( i = 0; i < idx_total; i++ )
464             if( srcb[i*step] )
465                 *dsti++ = i;
466     }
467     else
468     {
469         for( i = 0; i < idx_total; i++ )
470             dsti[i] = srci[i*step];
471 
472         if( !is_sorted )
473             qsort( dsti, idx_total, sizeof(dsti[0]), icvCmpIntegers );
474 
475         if( dsti[0] < 0 || dsti[idx_total-1] >= data_arr_size )
476             CV_ERROR( CV_StsOutOfRange, "the index array elements are out of range" );
477 
478         if( check_for_duplicates )
479         {
480             for( i = 1; i < idx_total; i++ )
481                 if( dsti[i] <= dsti[i-1] )
482                     CV_ERROR( CV_StsBadArg, "There are duplicated index array elements" );
483         }
484     }
485 
486     __END__;
487 
488     if( cvGetErrStatus() < 0 )
489         cvReleaseMat( &idx );
490 
491     return idx;
492 }
493 
494 
495 CvMat*
cvPreprocessVarType(const CvMat * var_type,const CvMat * var_idx,int var_count,int * response_type)496 cvPreprocessVarType( const CvMat* var_type, const CvMat* var_idx,
497                      int var_count, int* response_type )
498 {
499     CvMat* out_var_type = 0;
500     CV_FUNCNAME( "cvPreprocessVarType" );
501 
502     if( response_type )
503         *response_type = -1;
504 
505     __BEGIN__;
506 
507     int i, tm_size, tm_step;
508     //int* map = 0;
509     const uchar* src;
510     uchar* dst;
511 
512     if( !CV_IS_MAT(var_type) )
513         CV_ERROR( var_type ? CV_StsBadArg : CV_StsNullPtr, "Invalid or absent var_type array" );
514 
515     if( var_type->rows != 1 && var_type->cols != 1 )
516         CV_ERROR( CV_StsBadSize, "var_type array must be 1-dimensional" );
517 
518     if( !CV_IS_MASK_ARR(var_type))
519         CV_ERROR( CV_StsUnsupportedFormat, "type mask must be 8uC1 or 8sC1 array" );
520 
521     tm_size = var_type->rows + var_type->cols - 1;
522     tm_step = var_type->rows == 1 ? 1 : var_type->step/CV_ELEM_SIZE(var_type->type);
523 
524     if( /*tm_size != var_count &&*/ tm_size != var_count + 1 )
525         CV_ERROR( CV_StsBadArg,
526         "type mask must be of <input var count> + 1 size" );
527 
528     if( response_type && tm_size > var_count )
529         *response_type = var_type->data.ptr[var_count*tm_step] != 0;
530 
531     if( var_idx )
532     {
533         if( !CV_IS_MAT(var_idx) || CV_MAT_TYPE(var_idx->type) != CV_32SC1 ||
534             (var_idx->rows != 1 && var_idx->cols != 1) || !CV_IS_MAT_CONT(var_idx->type) )
535             CV_ERROR( CV_StsBadArg, "var index array should be continuous 1-dimensional integer vector" );
536         if( var_idx->rows + var_idx->cols - 1 > var_count )
537             CV_ERROR( CV_StsBadSize, "var index array is too large" );
538         //map = var_idx->data.i;
539         var_count = var_idx->rows + var_idx->cols - 1;
540     }
541 
542     CV_CALL( out_var_type = cvCreateMat( 1, var_count, CV_8UC1 ));
543     src = var_type->data.ptr;
544     dst = out_var_type->data.ptr;
545 
546     for( i = 0; i < var_count; i++ )
547     {
548         //int idx = map ? map[i] : i;
549         assert( (unsigned)/*idx*/i < (unsigned)tm_size );
550         dst[i] = (uchar)(src[/*idx*/i*tm_step] != 0);
551     }
552 
553     __END__;
554 
555     return out_var_type;
556 }
557 
558 
559 CvMat*
cvPreprocessOrderedResponses(const CvMat * responses,const CvMat * sample_idx,int sample_all)560 cvPreprocessOrderedResponses( const CvMat* responses, const CvMat* sample_idx, int sample_all )
561 {
562     CvMat* out_responses = 0;
563 
564     CV_FUNCNAME( "cvPreprocessOrderedResponses" );
565 
566     __BEGIN__;
567 
568     int i, r_type, r_step;
569     const int* map = 0;
570     float* dst;
571     int sample_count = sample_all;
572 
573     if( !CV_IS_MAT(responses) )
574         CV_ERROR( CV_StsBadArg, "Invalid response array" );
575 
576     if( responses->rows != 1 && responses->cols != 1 )
577         CV_ERROR( CV_StsBadSize, "Response array must be 1-dimensional" );
578 
579     if( responses->rows + responses->cols - 1 != sample_count )
580         CV_ERROR( CV_StsUnmatchedSizes,
581         "Response array must contain as many elements as the total number of samples" );
582 
583     r_type = CV_MAT_TYPE(responses->type);
584     if( r_type != CV_32FC1 && r_type != CV_32SC1 )
585         CV_ERROR( CV_StsUnsupportedFormat, "Unsupported response type" );
586 
587     r_step = responses->step ? responses->step / CV_ELEM_SIZE(responses->type) : 1;
588 
589     if( r_type == CV_32FC1 && CV_IS_MAT_CONT(responses->type) && !sample_idx )
590     {
591         out_responses = cvCloneMat( responses );
592         EXIT;
593     }
594 
595     if( sample_idx )
596     {
597         if( !CV_IS_MAT(sample_idx) || CV_MAT_TYPE(sample_idx->type) != CV_32SC1 ||
598             (sample_idx->rows != 1 && sample_idx->cols != 1) || !CV_IS_MAT_CONT(sample_idx->type) )
599             CV_ERROR( CV_StsBadArg, "sample index array should be continuous 1-dimensional integer vector" );
600         if( sample_idx->rows + sample_idx->cols - 1 > sample_count )
601             CV_ERROR( CV_StsBadSize, "sample index array is too large" );
602         map = sample_idx->data.i;
603         sample_count = sample_idx->rows + sample_idx->cols - 1;
604     }
605 
606     CV_CALL( out_responses = cvCreateMat( 1, sample_count, CV_32FC1 ));
607 
608     dst = out_responses->data.fl;
609     if( r_type == CV_32FC1 )
610     {
611         const float* src = responses->data.fl;
612         for( i = 0; i < sample_count; i++ )
613         {
614             int idx = map ? map[i] : i;
615             assert( (unsigned)idx < (unsigned)sample_all );
616             dst[i] = src[idx*r_step];
617         }
618     }
619     else
620     {
621         const int* src = responses->data.i;
622         for( i = 0; i < sample_count; i++ )
623         {
624             int idx = map ? map[i] : i;
625             assert( (unsigned)idx < (unsigned)sample_all );
626             dst[i] = (float)src[idx*r_step];
627         }
628     }
629 
630     __END__;
631 
632     return out_responses;
633 }
634 
635 CvMat*
cvPreprocessCategoricalResponses(const CvMat * responses,const CvMat * sample_idx,int sample_all,CvMat ** out_response_map,CvMat ** class_counts)636 cvPreprocessCategoricalResponses( const CvMat* responses,
637     const CvMat* sample_idx, int sample_all,
638     CvMat** out_response_map, CvMat** class_counts )
639 {
640     CvMat* out_responses = 0;
641     int** response_ptr = 0;
642 
643     CV_FUNCNAME( "cvPreprocessCategoricalResponses" );
644 
645     if( out_response_map )
646         *out_response_map = 0;
647 
648     if( class_counts )
649         *class_counts = 0;
650 
651     __BEGIN__;
652 
653     int i, r_type, r_step;
654     int cls_count = 1, prev_cls, prev_i;
655     const int* map = 0;
656     const int* srci;
657     const float* srcfl;
658     int* dst;
659     int* cls_map;
660     int* cls_counts = 0;
661     int sample_count = sample_all;
662 
663     if( !CV_IS_MAT(responses) )
664         CV_ERROR( CV_StsBadArg, "Invalid response array" );
665 
666     if( responses->rows != 1 && responses->cols != 1 )
667         CV_ERROR( CV_StsBadSize, "Response array must be 1-dimensional" );
668 
669     if( responses->rows + responses->cols - 1 != sample_count )
670         CV_ERROR( CV_StsUnmatchedSizes,
671         "Response array must contain as many elements as the total number of samples" );
672 
673     r_type = CV_MAT_TYPE(responses->type);
674     if( r_type != CV_32FC1 && r_type != CV_32SC1 )
675         CV_ERROR( CV_StsUnsupportedFormat, "Unsupported response type" );
676 
677     r_step = responses->rows == 1 ? 1 : responses->step / CV_ELEM_SIZE(responses->type);
678 
679     if( sample_idx )
680     {
681         if( !CV_IS_MAT(sample_idx) || CV_MAT_TYPE(sample_idx->type) != CV_32SC1 ||
682             (sample_idx->rows != 1 && sample_idx->cols != 1) || !CV_IS_MAT_CONT(sample_idx->type) )
683             CV_ERROR( CV_StsBadArg, "sample index array should be continuous 1-dimensional integer vector" );
684         if( sample_idx->rows + sample_idx->cols - 1 > sample_count )
685             CV_ERROR( CV_StsBadSize, "sample index array is too large" );
686         map = sample_idx->data.i;
687         sample_count = sample_idx->rows + sample_idx->cols - 1;
688     }
689 
690     CV_CALL( out_responses = cvCreateMat( 1, sample_count, CV_32SC1 ));
691 
692     if( !out_response_map )
693         CV_ERROR( CV_StsNullPtr, "out_response_map pointer is NULL" );
694 
695     CV_CALL( response_ptr = (int**)cvAlloc( sample_count*sizeof(response_ptr[0])));
696 
697     srci = responses->data.i;
698     srcfl = responses->data.fl;
699     dst = out_responses->data.i;
700 
701     for( i = 0; i < sample_count; i++ )
702     {
703         int idx = map ? map[i] : i;
704         assert( (unsigned)idx < (unsigned)sample_all );
705         if( r_type == CV_32SC1 )
706             dst[i] = srci[idx*r_step];
707         else
708         {
709             float rf = srcfl[idx*r_step];
710             int ri = cvRound(rf);
711             if( ri != rf )
712             {
713                 char buf[100];
714                 sprintf( buf, "response #%d is not integral", idx );
715                 CV_ERROR( CV_StsBadArg, buf );
716             }
717             dst[i] = ri;
718         }
719         response_ptr[i] = dst + i;
720     }
721 
722     qsort( response_ptr, sample_count, sizeof(int*), icvCmpIntegersPtr );
723 
724     // count the classes
725     for( i = 1; i < sample_count; i++ )
726         cls_count += *response_ptr[i] != *response_ptr[i-1];
727 
728     if( cls_count < 2 )
729         CV_ERROR( CV_StsBadArg, "There is only a single class" );
730 
731     CV_CALL( *out_response_map = cvCreateMat( 1, cls_count, CV_32SC1 ));
732 
733     if( class_counts )
734     {
735         CV_CALL( *class_counts = cvCreateMat( 1, cls_count, CV_32SC1 ));
736         cls_counts = (*class_counts)->data.i;
737     }
738 
739     // compact the class indices and build the map
740     prev_cls = ~*response_ptr[0];
741     cls_count = -1;
742     cls_map = (*out_response_map)->data.i;
743 
744     for( i = 0, prev_i = -1; i < sample_count; i++ )
745     {
746         int cur_cls = *response_ptr[i];
747         if( cur_cls != prev_cls )
748         {
749             if( cls_counts && cls_count >= 0 )
750                 cls_counts[cls_count] = i - prev_i;
751             cls_map[++cls_count] = prev_cls = cur_cls;
752             prev_i = i;
753         }
754         *response_ptr[i] = cls_count;
755     }
756 
757     if( cls_counts )
758         cls_counts[cls_count] = i - prev_i;
759 
760     __END__;
761 
762     cvFree( &response_ptr );
763 
764     return out_responses;
765 }
766 
767 
768 const float**
cvGetTrainSamples(const CvMat * train_data,int tflag,const CvMat * var_idx,const CvMat * sample_idx,int * _var_count,int * _sample_count,bool always_copy_data)769 cvGetTrainSamples( const CvMat* train_data, int tflag,
770                    const CvMat* var_idx, const CvMat* sample_idx,
771                    int* _var_count, int* _sample_count,
772                    bool always_copy_data )
773 {
774     float** samples = 0;
775 
776     CV_FUNCNAME( "cvGetTrainSamples" );
777 
778     __BEGIN__;
779 
780     int i, j, var_count, sample_count, s_step, v_step;
781     bool copy_data;
782     const float* data;
783     const int *s_idx, *v_idx;
784 
785     if( !CV_IS_MAT(train_data) )
786         CV_ERROR( CV_StsBadArg, "Invalid or NULL training data matrix" );
787 
788     var_count = var_idx ? var_idx->cols + var_idx->rows - 1 :
789                 tflag == CV_ROW_SAMPLE ? train_data->cols : train_data->rows;
790     sample_count = sample_idx ? sample_idx->cols + sample_idx->rows - 1 :
791                    tflag == CV_ROW_SAMPLE ? train_data->rows : train_data->cols;
792 
793     if( _var_count )
794         *_var_count = var_count;
795 
796     if( _sample_count )
797         *_sample_count = sample_count;
798 
799     copy_data = tflag != CV_ROW_SAMPLE || var_idx || always_copy_data;
800 
801     CV_CALL( samples = (float**)cvAlloc(sample_count*sizeof(samples[0]) +
802                 (copy_data ? 1 : 0)*var_count*sample_count*sizeof(samples[0][0])) );
803     data = train_data->data.fl;
804     s_step = train_data->step / sizeof(samples[0][0]);
805     v_step = 1;
806     s_idx = sample_idx ? sample_idx->data.i : 0;
807     v_idx = var_idx ? var_idx->data.i : 0;
808 
809     if( !copy_data )
810     {
811         for( i = 0; i < sample_count; i++ )
812             samples[i] = (float*)(data + (s_idx ? s_idx[i] : i)*s_step);
813     }
814     else
815     {
816         samples[0] = (float*)(samples + sample_count);
817         if( tflag != CV_ROW_SAMPLE )
818             CV_SWAP( s_step, v_step, i );
819 
820         for( i = 0; i < sample_count; i++ )
821         {
822             float* dst = samples[i] = samples[0] + i*var_count;
823             const float* src = data + (s_idx ? s_idx[i] : i)*s_step;
824 
825             if( !v_idx )
826                 for( j = 0; j < var_count; j++ )
827                     dst[j] = src[j*v_step];
828             else
829                 for( j = 0; j < var_count; j++ )
830                     dst[j] = src[v_idx[j]*v_step];
831         }
832     }
833 
834     __END__;
835 
836     return (const float**)samples;
837 }
838 
839 
840 void
cvCheckTrainData(const CvMat * train_data,int tflag,const CvMat * missing_mask,int * var_all,int * sample_all)841 cvCheckTrainData( const CvMat* train_data, int tflag,
842                   const CvMat* missing_mask,
843                   int* var_all, int* sample_all )
844 {
845     CV_FUNCNAME( "cvCheckTrainData" );
846 
847     if( var_all )
848         *var_all = 0;
849 
850     if( sample_all )
851         *sample_all = 0;
852 
853     __BEGIN__;
854 
855     // check parameter types and sizes
856     if( !CV_IS_MAT(train_data) || CV_MAT_TYPE(train_data->type) != CV_32FC1 )
857         CV_ERROR( CV_StsBadArg, "train data must be floating-point matrix" );
858 
859     if( missing_mask )
860     {
861         if( !CV_IS_MAT(missing_mask) || !CV_IS_MASK_ARR(missing_mask) ||
862             !CV_ARE_SIZES_EQ(train_data, missing_mask) )
863             CV_ERROR( CV_StsBadArg,
864             "missing value mask must be 8-bit matrix of the same size as training data" );
865     }
866 
867     if( tflag != CV_ROW_SAMPLE && tflag != CV_COL_SAMPLE )
868         CV_ERROR( CV_StsBadArg,
869         "Unknown training data layout (must be CV_ROW_SAMPLE or CV_COL_SAMPLE)" );
870 
871     if( var_all )
872         *var_all = tflag == CV_ROW_SAMPLE ? train_data->cols : train_data->rows;
873 
874     if( sample_all )
875         *sample_all = tflag == CV_ROW_SAMPLE ? train_data->rows : train_data->cols;
876 
877     __END__;
878 }
879 
880 
881 int
cvPrepareTrainData(const char *,const CvMat * train_data,int tflag,const CvMat * responses,int response_type,const CvMat * var_idx,const CvMat * sample_idx,bool always_copy_data,const float *** out_train_samples,int * _sample_count,int * _var_count,int * _var_all,CvMat ** out_responses,CvMat ** out_response_map,CvMat ** out_var_idx,CvMat ** out_sample_idx)882 cvPrepareTrainData( const char* /*funcname*/,
883                     const CvMat* train_data, int tflag,
884                     const CvMat* responses, int response_type,
885                     const CvMat* var_idx,
886                     const CvMat* sample_idx,
887                     bool always_copy_data,
888                     const float*** out_train_samples,
889                     int* _sample_count,
890                     int* _var_count,
891                     int* _var_all,
892                     CvMat** out_responses,
893                     CvMat** out_response_map,
894                     CvMat** out_var_idx,
895                     CvMat** out_sample_idx )
896 {
897     int ok = 0;
898     CvMat* _var_idx = 0;
899     CvMat* _sample_idx = 0;
900     CvMat* _responses = 0;
901     int sample_all = 0, sample_count = 0, var_all = 0, var_count = 0;
902 
903     CV_FUNCNAME( "cvPrepareTrainData" );
904 
905     // step 0. clear all the output pointers to ensure we do not try
906     // to call free() with uninitialized pointers
907     if( out_responses )
908         *out_responses = 0;
909 
910     if( out_response_map )
911         *out_response_map = 0;
912 
913     if( out_var_idx )
914         *out_var_idx = 0;
915 
916     if( out_sample_idx )
917         *out_sample_idx = 0;
918 
919     if( out_train_samples )
920         *out_train_samples = 0;
921 
922     if( _sample_count )
923         *_sample_count = 0;
924 
925     if( _var_count )
926         *_var_count = 0;
927 
928     if( _var_all )
929         *_var_all = 0;
930 
931     __BEGIN__;
932 
933     if( !out_train_samples )
934         CV_ERROR( CV_StsBadArg, "output pointer to train samples is NULL" );
935 
936     CV_CALL( cvCheckTrainData( train_data, tflag, 0, &var_all, &sample_all ));
937 
938     if( sample_idx )
939         CV_CALL( _sample_idx = cvPreprocessIndexArray( sample_idx, sample_all ));
940     if( var_idx )
941         CV_CALL( _var_idx = cvPreprocessIndexArray( var_idx, var_all ));
942 
943     if( responses )
944     {
945         if( !out_responses )
946             CV_ERROR( CV_StsNullPtr, "output response pointer is NULL" );
947 
948         if( response_type == CV_VAR_NUMERICAL )
949         {
950             CV_CALL( _responses = cvPreprocessOrderedResponses( responses,
951                                                 _sample_idx, sample_all ));
952         }
953         else
954         {
955             CV_CALL( _responses = cvPreprocessCategoricalResponses( responses,
956                                 _sample_idx, sample_all, out_response_map, 0 ));
957         }
958     }
959 
960     CV_CALL( *out_train_samples =
961                 cvGetTrainSamples( train_data, tflag, _var_idx, _sample_idx,
962                                    &var_count, &sample_count, always_copy_data ));
963 
964     ok = 1;
965 
966     __END__;
967 
968     if( ok )
969     {
970         if( out_responses )
971             *out_responses = _responses, _responses = 0;
972 
973         if( out_var_idx )
974             *out_var_idx = _var_idx, _var_idx = 0;
975 
976         if( out_sample_idx )
977             *out_sample_idx = _sample_idx, _sample_idx = 0;
978 
979         if( _sample_count )
980             *_sample_count = sample_count;
981 
982         if( _var_count )
983             *_var_count = var_count;
984 
985         if( _var_all )
986             *_var_all = var_all;
987     }
988     else
989     {
990         if( out_response_map )
991             cvReleaseMat( out_response_map );
992         cvFree( out_train_samples );
993     }
994 
995     if( _responses != responses )
996         cvReleaseMat( &_responses );
997     cvReleaseMat( &_var_idx );
998     cvReleaseMat( &_sample_idx );
999 
1000     return ok;
1001 }
1002 
1003 
1004 typedef struct CvSampleResponsePair
1005 {
1006     const float* sample;
1007     const uchar* mask;
1008     int response;
1009     int index;
1010 }
1011 CvSampleResponsePair;
1012 
1013 
1014 static int
icvCmpSampleResponsePairs(const void * a,const void * b)1015 CV_CDECL icvCmpSampleResponsePairs( const void* a, const void* b )
1016 {
1017     int ra = ((const CvSampleResponsePair*)a)->response;
1018     int rb = ((const CvSampleResponsePair*)b)->response;
1019     int ia = ((const CvSampleResponsePair*)a)->index;
1020     int ib = ((const CvSampleResponsePair*)b)->index;
1021 
1022     return ra < rb ? -1 : ra > rb ? 1 : ia - ib;
1023     //return (ra > rb ? -1 : 0)|(ra < rb);
1024 }
1025 
1026 
1027 void
cvSortSamplesByClasses(const float ** samples,const CvMat * classes,int * class_ranges,const uchar ** mask)1028 cvSortSamplesByClasses( const float** samples, const CvMat* classes,
1029                         int* class_ranges, const uchar** mask )
1030 {
1031     CvSampleResponsePair* pairs = 0;
1032     CV_FUNCNAME( "cvSortSamplesByClasses" );
1033 
1034     __BEGIN__;
1035 
1036     int i, k = 0, sample_count;
1037 
1038     if( !samples || !classes || !class_ranges )
1039         CV_ERROR( CV_StsNullPtr, "INTERNAL ERROR: some of the args are NULL pointers" );
1040 
1041     if( classes->rows != 1 || CV_MAT_TYPE(classes->type) != CV_32SC1 )
1042         CV_ERROR( CV_StsBadArg, "classes array must be a single row of integers" );
1043 
1044     sample_count = classes->cols;
1045     CV_CALL( pairs = (CvSampleResponsePair*)cvAlloc( (sample_count+1)*sizeof(pairs[0])));
1046 
1047     for( i = 0; i < sample_count; i++ )
1048     {
1049         pairs[i].sample = samples[i];
1050         pairs[i].mask = (mask) ? (mask[i]) : 0;
1051         pairs[i].response = classes->data.i[i];
1052         pairs[i].index = i;
1053         assert( classes->data.i[i] >= 0 );
1054     }
1055 
1056     qsort( pairs, sample_count, sizeof(pairs[0]), icvCmpSampleResponsePairs );
1057     pairs[sample_count].response = -1;
1058     class_ranges[0] = 0;
1059 
1060     for( i = 0; i < sample_count; i++ )
1061     {
1062         samples[i] = pairs[i].sample;
1063         if (mask)
1064             mask[i] = pairs[i].mask;
1065         classes->data.i[i] = pairs[i].response;
1066 
1067         if( pairs[i].response != pairs[i+1].response )
1068             class_ranges[++k] = i+1;
1069     }
1070 
1071     __END__;
1072 
1073     cvFree( &pairs );
1074 }
1075 
1076 
1077 void
cvPreparePredictData(const CvArr * _sample,int dims_all,const CvMat * comp_idx,int class_count,const CvMat * prob,float ** _row_sample,int as_sparse)1078 cvPreparePredictData( const CvArr* _sample, int dims_all,
1079                       const CvMat* comp_idx, int class_count,
1080                       const CvMat* prob, float** _row_sample,
1081                       int as_sparse )
1082 {
1083     float* row_sample = 0;
1084     int* inverse_comp_idx = 0;
1085 
1086     CV_FUNCNAME( "cvPreparePredictData" );
1087 
1088     __BEGIN__;
1089 
1090     const CvMat* sample = (const CvMat*)_sample;
1091     float* sample_data;
1092     int sample_step;
1093     int is_sparse = CV_IS_SPARSE_MAT(sample);
1094     int d, sizes[CV_MAX_DIM];
1095     int i, dims_selected;
1096     int vec_size;
1097 
1098     if( !is_sparse && !CV_IS_MAT(sample) )
1099         CV_ERROR( !sample ? CV_StsNullPtr : CV_StsBadArg, "The sample is not a valid vector" );
1100 
1101     if( cvGetElemType( sample ) != CV_32FC1 )
1102         CV_ERROR( CV_StsUnsupportedFormat, "Input sample must have 32fC1 type" );
1103 
1104     CV_CALL( d = cvGetDims( sample, sizes ));
1105 
1106     if( !((is_sparse && d == 1) || (!is_sparse && d == 2 && (sample->rows == 1 || sample->cols == 1))) )
1107         CV_ERROR( CV_StsBadSize, "Input sample must be 1-dimensional vector" );
1108 
1109     if( d == 1 )
1110         sizes[1] = 1;
1111 
1112     if( sizes[0] + sizes[1] - 1 != dims_all )
1113         CV_ERROR( CV_StsUnmatchedSizes,
1114         "The sample size is different from what has been used for training" );
1115 
1116     if( !_row_sample )
1117         CV_ERROR( CV_StsNullPtr, "INTERNAL ERROR: The row_sample pointer is NULL" );
1118 
1119     if( comp_idx && (!CV_IS_MAT(comp_idx) || comp_idx->rows != 1 ||
1120         CV_MAT_TYPE(comp_idx->type) != CV_32SC1) )
1121         CV_ERROR( CV_StsBadArg, "INTERNAL ERROR: invalid comp_idx" );
1122 
1123     dims_selected = comp_idx ? comp_idx->cols : dims_all;
1124 
1125     if( prob )
1126     {
1127         if( !CV_IS_MAT(prob) )
1128             CV_ERROR( CV_StsBadArg, "The output matrix of probabilities is invalid" );
1129 
1130         if( (prob->rows != 1 && prob->cols != 1) ||
1131             (CV_MAT_TYPE(prob->type) != CV_32FC1 &&
1132             CV_MAT_TYPE(prob->type) != CV_64FC1) )
1133             CV_ERROR( CV_StsBadSize,
1134             "The matrix of probabilities must be 1-dimensional vector of 32fC1 type" );
1135 
1136         if( prob->rows + prob->cols - 1 != class_count )
1137             CV_ERROR( CV_StsUnmatchedSizes,
1138             "The vector of probabilities must contain as many elements as "
1139             "the number of classes in the training set" );
1140     }
1141 
1142     vec_size = !as_sparse ? dims_selected*sizeof(row_sample[0]) :
1143                 (dims_selected + 1)*sizeof(CvSparseVecElem32f);
1144 
1145     if( CV_IS_MAT(sample) )
1146     {
1147         sample_data = sample->data.fl;
1148         sample_step = CV_IS_MAT_CONT(sample->type) ? 1 : sample->step/sizeof(row_sample[0]);
1149 
1150         if( !comp_idx && CV_IS_MAT_CONT(sample->type) && !as_sparse )
1151             *_row_sample = sample_data;
1152         else
1153         {
1154             CV_CALL( row_sample = (float*)cvAlloc( vec_size ));
1155 
1156             if( !comp_idx )
1157                 for( i = 0; i < dims_selected; i++ )
1158                     row_sample[i] = sample_data[sample_step*i];
1159             else
1160             {
1161                 int* comp = comp_idx->data.i;
1162                 for( i = 0; i < dims_selected; i++ )
1163                     row_sample[i] = sample_data[sample_step*comp[i]];
1164             }
1165 
1166             *_row_sample = row_sample;
1167         }
1168 
1169         if( as_sparse )
1170         {
1171             const float* src = (const float*)row_sample;
1172             CvSparseVecElem32f* dst = (CvSparseVecElem32f*)row_sample;
1173 
1174             dst[dims_selected].idx = -1;
1175             for( i = dims_selected - 1; i >= 0; i-- )
1176             {
1177                 dst[i].idx = i;
1178                 dst[i].val = src[i];
1179             }
1180         }
1181     }
1182     else
1183     {
1184         CvSparseNode* node;
1185         CvSparseMatIterator mat_iterator;
1186         const CvSparseMat* sparse = (const CvSparseMat*)sample;
1187         assert( is_sparse );
1188 
1189         node = cvInitSparseMatIterator( sparse, &mat_iterator );
1190         CV_CALL( row_sample = (float*)cvAlloc( vec_size ));
1191 
1192         if( comp_idx )
1193         {
1194             CV_CALL( inverse_comp_idx = (int*)cvAlloc( dims_all*sizeof(int) ));
1195             memset( inverse_comp_idx, -1, dims_all*sizeof(int) );
1196             for( i = 0; i < dims_selected; i++ )
1197                 inverse_comp_idx[comp_idx->data.i[i]] = i;
1198         }
1199 
1200         if( !as_sparse )
1201         {
1202             memset( row_sample, 0, vec_size );
1203 
1204             for( ; node != 0; node = cvGetNextSparseNode(&mat_iterator) )
1205             {
1206                 int idx = *CV_NODE_IDX( sparse, node );
1207                 if( inverse_comp_idx )
1208                 {
1209                     idx = inverse_comp_idx[idx];
1210                     if( idx < 0 )
1211                         continue;
1212                 }
1213                 row_sample[idx] = *(float*)CV_NODE_VAL( sparse, node );
1214             }
1215         }
1216         else
1217         {
1218             CvSparseVecElem32f* ptr = (CvSparseVecElem32f*)row_sample;
1219 
1220             for( ; node != 0; node = cvGetNextSparseNode(&mat_iterator) )
1221             {
1222                 int idx = *CV_NODE_IDX( sparse, node );
1223                 if( inverse_comp_idx )
1224                 {
1225                     idx = inverse_comp_idx[idx];
1226                     if( idx < 0 )
1227                         continue;
1228                 }
1229                 ptr->idx = idx;
1230                 ptr->val = *(float*)CV_NODE_VAL( sparse, node );
1231                 ptr++;
1232             }
1233 
1234             qsort( row_sample, ptr - (CvSparseVecElem32f*)row_sample,
1235                    sizeof(ptr[0]), icvCmpSparseVecElems );
1236             ptr->idx = -1;
1237         }
1238 
1239         *_row_sample = row_sample;
1240     }
1241 
1242     __END__;
1243 
1244     if( inverse_comp_idx )
1245         cvFree( &inverse_comp_idx );
1246 
1247     if( cvGetErrStatus() < 0 && _row_sample )
1248     {
1249         cvFree( &row_sample );
1250         *_row_sample = 0;
1251     }
1252 }
1253 
1254 
1255 static void
icvConvertDataToSparse(const uchar * src,int src_step,int src_type,uchar * dst,int dst_step,int dst_type,CvSize size,int * idx)1256 icvConvertDataToSparse( const uchar* src, int src_step, int src_type,
1257                         uchar* dst, int dst_step, int dst_type,
1258                         CvSize size, int* idx )
1259 {
1260     CV_FUNCNAME( "icvConvertDataToSparse" );
1261 
1262     __BEGIN__;
1263 
1264     int i, j;
1265     src_type = CV_MAT_TYPE(src_type);
1266     dst_type = CV_MAT_TYPE(dst_type);
1267 
1268     if( CV_MAT_CN(src_type) != 1 || CV_MAT_CN(dst_type) != 1 )
1269         CV_ERROR( CV_StsUnsupportedFormat, "The function supports only single-channel arrays" );
1270 
1271     if( src_step == 0 )
1272         src_step = CV_ELEM_SIZE(src_type);
1273 
1274     if( dst_step == 0 )
1275         dst_step = CV_ELEM_SIZE(dst_type);
1276 
1277     // if there is no "idx" and if both arrays are continuous,
1278     // do the whole processing (copying or conversion) in a single loop
1279     if( !idx && CV_ELEM_SIZE(src_type)*size.width == src_step &&
1280         CV_ELEM_SIZE(dst_type)*size.width == dst_step )
1281     {
1282         size.width *= size.height;
1283         size.height = 1;
1284     }
1285 
1286     if( src_type == dst_type )
1287     {
1288         int full_width = CV_ELEM_SIZE(dst_type)*size.width;
1289 
1290         if( full_width == sizeof(int) ) // another common case: copy int's or float's
1291             for( i = 0; i < size.height; i++, src += src_step )
1292                 *(int*)(dst + dst_step*(idx ? idx[i] : i)) = *(int*)src;
1293         else
1294             for( i = 0; i < size.height; i++, src += src_step )
1295                 memcpy( dst + dst_step*(idx ? idx[i] : i), src, full_width );
1296     }
1297     else if( src_type == CV_32SC1 && (dst_type == CV_32FC1 || dst_type == CV_64FC1) )
1298         for( i = 0; i < size.height; i++, src += src_step )
1299         {
1300             uchar* _dst = dst + dst_step*(idx ? idx[i] : i);
1301             if( dst_type == CV_32FC1 )
1302                 for( j = 0; j < size.width; j++ )
1303                     ((float*)_dst)[j] = (float)((int*)src)[j];
1304             else
1305                 for( j = 0; j < size.width; j++ )
1306                     ((double*)_dst)[j] = ((int*)src)[j];
1307         }
1308     else if( (src_type == CV_32FC1 || src_type == CV_64FC1) && dst_type == CV_32SC1 )
1309         for( i = 0; i < size.height; i++, src += src_step )
1310         {
1311             uchar* _dst = dst + dst_step*(idx ? idx[i] : i);
1312             if( src_type == CV_32FC1 )
1313                 for( j = 0; j < size.width; j++ )
1314                     ((int*)_dst)[j] = cvRound(((float*)src)[j]);
1315             else
1316                 for( j = 0; j < size.width; j++ )
1317                     ((int*)_dst)[j] = cvRound(((double*)src)[j]);
1318         }
1319     else if( (src_type == CV_32FC1 && dst_type == CV_64FC1) ||
1320              (src_type == CV_64FC1 && dst_type == CV_32FC1) )
1321         for( i = 0; i < size.height; i++, src += src_step )
1322         {
1323             uchar* _dst = dst + dst_step*(idx ? idx[i] : i);
1324             if( src_type == CV_32FC1 )
1325                 for( j = 0; j < size.width; j++ )
1326                     ((double*)_dst)[j] = ((float*)src)[j];
1327             else
1328                 for( j = 0; j < size.width; j++ )
1329                     ((float*)_dst)[j] = (float)((double*)src)[j];
1330         }
1331     else
1332         CV_ERROR( CV_StsUnsupportedFormat, "Unsupported combination of input and output vectors" );
1333 
1334     __END__;
1335 }
1336 
1337 
1338 void
cvWritebackLabels(const CvMat * labels,CvMat * dst_labels,const CvMat * centers,CvMat * dst_centers,const CvMat * probs,CvMat * dst_probs,const CvMat * sample_idx,int samples_all,const CvMat * comp_idx,int dims_all)1339 cvWritebackLabels( const CvMat* labels, CvMat* dst_labels,
1340                    const CvMat* centers, CvMat* dst_centers,
1341                    const CvMat* probs, CvMat* dst_probs,
1342                    const CvMat* sample_idx, int samples_all,
1343                    const CvMat* comp_idx, int dims_all )
1344 {
1345     CV_FUNCNAME( "cvWritebackLabels" );
1346 
1347     __BEGIN__;
1348 
1349     int samples_selected = samples_all, dims_selected = dims_all;
1350 
1351     if( dst_labels && !CV_IS_MAT(dst_labels) )
1352         CV_ERROR( CV_StsBadArg, "Array of output labels is not a valid matrix" );
1353 
1354     if( dst_centers )
1355         if( !ICV_IS_MAT_OF_TYPE(dst_centers, CV_32FC1) &&
1356             !ICV_IS_MAT_OF_TYPE(dst_centers, CV_64FC1) )
1357             CV_ERROR( CV_StsBadArg, "Array of cluster centers is not a valid matrix" );
1358 
1359     if( dst_probs && !CV_IS_MAT(dst_probs) )
1360         CV_ERROR( CV_StsBadArg, "Probability matrix is not valid" );
1361 
1362     if( sample_idx )
1363     {
1364         CV_ASSERT( sample_idx->rows == 1 && CV_MAT_TYPE(sample_idx->type) == CV_32SC1 );
1365         samples_selected = sample_idx->cols;
1366     }
1367 
1368     if( comp_idx )
1369     {
1370         CV_ASSERT( comp_idx->rows == 1 && CV_MAT_TYPE(comp_idx->type) == CV_32SC1 );
1371         dims_selected = comp_idx->cols;
1372     }
1373 
1374     if( dst_labels && (!labels || labels->data.ptr != dst_labels->data.ptr) )
1375     {
1376         if( !labels )
1377             CV_ERROR( CV_StsNullPtr, "NULL labels" );
1378 
1379         CV_ASSERT( labels->rows == 1 );
1380 
1381         if( dst_labels->rows != 1 && dst_labels->cols != 1 )
1382             CV_ERROR( CV_StsBadSize, "Array of output labels should be 1d vector" );
1383 
1384         if( dst_labels->rows + dst_labels->cols - 1 != samples_all )
1385             CV_ERROR( CV_StsUnmatchedSizes,
1386             "Size of vector of output labels is not equal to the total number of input samples" );
1387 
1388         CV_ASSERT( labels->cols == samples_selected );
1389 
1390         CV_CALL( icvConvertDataToSparse( labels->data.ptr, labels->step, labels->type,
1391                         dst_labels->data.ptr, dst_labels->step, dst_labels->type,
1392                         cvSize( 1, samples_selected ), sample_idx ? sample_idx->data.i : 0 ));
1393     }
1394 
1395     if( dst_centers && (!centers || centers->data.ptr != dst_centers->data.ptr) )
1396     {
1397         int i;
1398 
1399         if( !centers )
1400             CV_ERROR( CV_StsNullPtr, "NULL centers" );
1401 
1402         if( centers->rows != dst_centers->rows )
1403             CV_ERROR( CV_StsUnmatchedSizes, "Invalid number of rows in matrix of output centers" );
1404 
1405         if( dst_centers->cols != dims_all )
1406             CV_ERROR( CV_StsUnmatchedSizes,
1407             "Number of columns in matrix of output centers is "
1408             "not equal to the total number of components in the input samples" );
1409 
1410         CV_ASSERT( centers->cols == dims_selected );
1411 
1412         for( i = 0; i < centers->rows; i++ )
1413             CV_CALL( icvConvertDataToSparse( centers->data.ptr + i*centers->step, 0, centers->type,
1414                         dst_centers->data.ptr + i*dst_centers->step, 0, dst_centers->type,
1415                         cvSize( 1, dims_selected ), comp_idx ? comp_idx->data.i : 0 ));
1416     }
1417 
1418     if( dst_probs && (!probs || probs->data.ptr != dst_probs->data.ptr) )
1419     {
1420         if( !probs )
1421             CV_ERROR( CV_StsNullPtr, "NULL probs" );
1422 
1423         if( probs->cols != dst_probs->cols )
1424             CV_ERROR( CV_StsUnmatchedSizes, "Invalid number of columns in output probability matrix" );
1425 
1426         if( dst_probs->rows != samples_all )
1427             CV_ERROR( CV_StsUnmatchedSizes,
1428             "Number of rows in output probability matrix is "
1429             "not equal to the total number of input samples" );
1430 
1431         CV_ASSERT( probs->rows == samples_selected );
1432 
1433         CV_CALL( icvConvertDataToSparse( probs->data.ptr, probs->step, probs->type,
1434                         dst_probs->data.ptr, dst_probs->step, dst_probs->type,
1435                         cvSize( probs->cols, samples_selected ),
1436                         sample_idx ? sample_idx->data.i : 0 ));
1437     }
1438 
1439     __END__;
1440 }
1441 
1442 #if 0
1443 CV_IMPL void
1444 cvStatModelMultiPredict( const CvStatModel* stat_model,
1445                          const CvArr* predict_input,
1446                          int flags, CvMat* predict_output,
1447                          CvMat* probs, const CvMat* sample_idx )
1448 {
1449     CvMemStorage* storage = 0;
1450     CvMat* sample_idx_buffer = 0;
1451     CvSparseMat** sparse_rows = 0;
1452     int samples_selected = 0;
1453 
1454     CV_FUNCNAME( "cvStatModelMultiPredict" );
1455 
1456     __BEGIN__;
1457 
1458     int i;
1459     int predict_output_step = 1, sample_idx_step = 1;
1460     int type;
1461     int d, sizes[CV_MAX_DIM];
1462     int tflag = flags == CV_COL_SAMPLE;
1463     int samples_all, dims_all;
1464     int is_sparse = CV_IS_SPARSE_MAT(predict_input);
1465     CvMat predict_input_part;
1466     CvArr* sample = &predict_input_part;
1467     CvMat probs_part;
1468     CvMat* probs1 = probs ? &probs_part : 0;
1469 
1470     if( !CV_IS_STAT_MODEL(stat_model) )
1471         CV_ERROR( !stat_model ? CV_StsNullPtr : CV_StsBadArg, "Invalid statistical model" );
1472 
1473     if( !stat_model->predict )
1474         CV_ERROR( CV_StsNotImplemented, "There is no \"predict\" method" );
1475 
1476     if( !predict_input || !predict_output )
1477         CV_ERROR( CV_StsNullPtr, "NULL input or output matrices" );
1478 
1479     if( !is_sparse && !CV_IS_MAT(predict_input) )
1480         CV_ERROR( CV_StsBadArg, "predict_input should be a matrix or a sparse matrix" );
1481 
1482     if( !CV_IS_MAT(predict_output) )
1483         CV_ERROR( CV_StsBadArg, "predict_output should be a matrix" );
1484 
1485     type = cvGetElemType( predict_input );
1486     if( type != CV_32FC1 ||
1487         (CV_MAT_TYPE(predict_output->type) != CV_32FC1 &&
1488          CV_MAT_TYPE(predict_output->type) != CV_32SC1 ))
1489          CV_ERROR( CV_StsUnsupportedFormat, "The input or output matrix has unsupported format" );
1490 
1491     CV_CALL( d = cvGetDims( predict_input, sizes ));
1492     if( d > 2 )
1493         CV_ERROR( CV_StsBadSize, "The input matrix should be 1- or 2-dimensional" );
1494 
1495     if( !tflag )
1496     {
1497         samples_all = samples_selected = sizes[0];
1498         dims_all = sizes[1];
1499     }
1500     else
1501     {
1502         samples_all = samples_selected = sizes[1];
1503         dims_all = sizes[0];
1504     }
1505 
1506     if( sample_idx )
1507     {
1508         if( !CV_IS_MAT(sample_idx) )
1509             CV_ERROR( CV_StsBadArg, "Invalid sample_idx matrix" );
1510 
1511         if( sample_idx->cols != 1 && sample_idx->rows != 1 )
1512             CV_ERROR( CV_StsBadSize, "sample_idx must be 1-dimensional matrix" );
1513 
1514         samples_selected = sample_idx->rows + sample_idx->cols - 1;
1515 
1516         if( CV_MAT_TYPE(sample_idx->type) == CV_32SC1 )
1517         {
1518             if( samples_selected > samples_all )
1519                 CV_ERROR( CV_StsBadSize, "sample_idx is too large vector" );
1520         }
1521         else if( samples_selected != samples_all )
1522             CV_ERROR( CV_StsUnmatchedSizes, "sample_idx has incorrect size" );
1523 
1524         sample_idx_step = sample_idx->step ?
1525             sample_idx->step / CV_ELEM_SIZE(sample_idx->type) : 1;
1526     }
1527 
1528     if( predict_output->rows != 1 && predict_output->cols != 1 )
1529         CV_ERROR( CV_StsBadSize, "predict_output should be a 1-dimensional matrix" );
1530 
1531     if( predict_output->rows + predict_output->cols - 1 != samples_all )
1532         CV_ERROR( CV_StsUnmatchedSizes, "predict_output and predict_input have uncoordinated sizes" );
1533 
1534     predict_output_step = predict_output->step ?
1535         predict_output->step / CV_ELEM_SIZE(predict_output->type) : 1;
1536 
1537     if( probs )
1538     {
1539         if( !CV_IS_MAT(probs) )
1540             CV_ERROR( CV_StsBadArg, "Invalid matrix of probabilities" );
1541 
1542         if( probs->rows != samples_all )
1543             CV_ERROR( CV_StsUnmatchedSizes,
1544             "matrix of probabilities must have as many rows as the total number of samples" );
1545 
1546         if( CV_MAT_TYPE(probs->type) != CV_32FC1 )
1547             CV_ERROR( CV_StsUnsupportedFormat, "matrix of probabilities must have 32fC1 type" );
1548     }
1549 
1550     if( is_sparse )
1551     {
1552         CvSparseNode* node;
1553         CvSparseMatIterator mat_iterator;
1554         CvSparseMat* sparse = (CvSparseMat*)predict_input;
1555 
1556         if( sample_idx && CV_MAT_TYPE(sample_idx->type) == CV_32SC1 )
1557         {
1558             CV_CALL( sample_idx_buffer = cvCreateMat( 1, samples_all, CV_8UC1 ));
1559             cvZero( sample_idx_buffer );
1560             for( i = 0; i < samples_selected; i++ )
1561                 sample_idx_buffer->data.ptr[sample_idx->data.i[i*sample_idx_step]] = 1;
1562             samples_selected = samples_all;
1563             sample_idx = sample_idx_buffer;
1564             sample_idx_step = 1;
1565         }
1566 
1567         CV_CALL( sparse_rows = (CvSparseMat**)cvAlloc( samples_selected*sizeof(sparse_rows[0])));
1568         for( i = 0; i < samples_selected; i++ )
1569         {
1570             if( sample_idx && sample_idx->data.ptr[i*sample_idx_step] == 0 )
1571                 continue;
1572             CV_CALL( sparse_rows[i] = cvCreateSparseMat( 1, &dims_all, type ));
1573             if( !storage )
1574                 storage = sparse_rows[i]->heap->storage;
1575             else
1576             {
1577                 // hack: to decrease memory footprint, make all the sparse matrices
1578                 // reside in the same storage
1579                 int elem_size = sparse_rows[i]->heap->elem_size;
1580                 cvReleaseMemStorage( &sparse_rows[i]->heap->storage );
1581                 sparse_rows[i]->heap = cvCreateSet( 0, sizeof(CvSet), elem_size, storage );
1582             }
1583         }
1584 
1585         // put each row (or column) of predict_input into separate sparse matrix.
1586         node = cvInitSparseMatIterator( sparse, &mat_iterator );
1587         for( ; node != 0; node = cvGetNextSparseNode( &mat_iterator ))
1588         {
1589             int* idx = CV_NODE_IDX( sparse, node );
1590             int idx0 = idx[tflag ^ 1];
1591             int idx1 = idx[tflag];
1592 
1593             if( sample_idx && sample_idx->data.ptr[idx0*sample_idx_step] == 0 )
1594                 continue;
1595 
1596             assert( sparse_rows[idx0] != 0 );
1597             *(float*)cvPtrND( sparse, &idx1, 0, 1, 0 ) = *(float*)CV_NODE_VAL( sparse, node );
1598         }
1599     }
1600 
1601     for( i = 0; i < samples_selected; i++ )
1602     {
1603         int idx = i;
1604         float response;
1605 
1606         if( sample_idx )
1607         {
1608             if( CV_MAT_TYPE(sample_idx->type) == CV_32SC1 )
1609             {
1610                 idx = sample_idx->data.i[i*sample_idx_step];
1611                 if( (unsigned)idx >= (unsigned)samples_all )
1612                     CV_ERROR( CV_StsOutOfRange, "Some of sample_idx elements are out of range" );
1613             }
1614             else if( CV_MAT_TYPE(sample_idx->type) == CV_8UC1 &&
1615                      sample_idx->data.ptr[i*sample_idx_step] == 0 )
1616                 continue;
1617         }
1618 
1619         if( !is_sparse )
1620         {
1621             if( !tflag )
1622                 cvGetRow( predict_input, &predict_input_part, idx );
1623             else
1624             {
1625                 cvGetCol( predict_input, &predict_input_part, idx );
1626             }
1627         }
1628         else
1629             sample = sparse_rows[idx];
1630 
1631         if( probs )
1632             cvGetRow( probs, probs1, idx );
1633 
1634         CV_CALL( response = stat_model->predict( stat_model, (CvMat*)sample, probs1 ));
1635 
1636         if( CV_MAT_TYPE(predict_output->type) == CV_32FC1 )
1637             predict_output->data.fl[idx*predict_output_step] = response;
1638         else
1639         {
1640             CV_ASSERT( cvRound(response) == response );
1641             predict_output->data.i[idx*predict_output_step] = cvRound(response);
1642         }
1643     }
1644 
1645     __END__;
1646 
1647     if( sparse_rows )
1648     {
1649         int i;
1650         for( i = 0; i < samples_selected; i++ )
1651             if( sparse_rows[i] )
1652             {
1653                 sparse_rows[i]->heap->storage = 0;
1654                 cvReleaseSparseMat( &sparse_rows[i] );
1655             }
1656         cvFree( &sparse_rows );
1657     }
1658 
1659     cvReleaseMat( &sample_idx_buffer );
1660     cvReleaseMemStorage( &storage );
1661 }
1662 #endif
1663 
1664 // By P. Yarykin - begin -
1665 
cvCombineResponseMaps(CvMat * _responses,const CvMat * old_response_map,CvMat * new_response_map,CvMat ** out_response_map)1666 void cvCombineResponseMaps (CvMat*  _responses,
1667                       const CvMat*  old_response_map,
1668                             CvMat*  new_response_map,
1669                             CvMat** out_response_map)
1670 {
1671     int** old_data = NULL;
1672     int** new_data = NULL;
1673 
1674         CV_FUNCNAME ("cvCombineResponseMaps");
1675         __BEGIN__
1676 
1677     int i,j;
1678     int old_n, new_n, out_n;
1679     int samples, free_response;
1680     int* first;
1681     int* responses;
1682     int* out_data;
1683 
1684     if( out_response_map )
1685         *out_response_map = 0;
1686 
1687 // Check input data.
1688     if ((!ICV_IS_MAT_OF_TYPE (_responses, CV_32SC1)) ||
1689         (!ICV_IS_MAT_OF_TYPE (old_response_map, CV_32SC1)) ||
1690         (!ICV_IS_MAT_OF_TYPE (new_response_map, CV_32SC1)))
1691     {
1692         CV_ERROR (CV_StsBadArg, "Some of input arguments is not the CvMat")
1693     }
1694 
1695 // Prepare sorted responses.
1696     first = new_response_map->data.i;
1697     new_n = new_response_map->cols;
1698     CV_CALL (new_data = (int**)cvAlloc (new_n * sizeof (new_data[0])));
1699     for (i = 0; i < new_n; i++)
1700         new_data[i] = first + i;
1701     qsort (new_data, new_n, sizeof(int*), icvCmpIntegersPtr);
1702 
1703     first = old_response_map->data.i;
1704     old_n = old_response_map->cols;
1705     CV_CALL (old_data = (int**)cvAlloc (old_n * sizeof (old_data[0])));
1706     for (i = 0; i < old_n; i++)
1707         old_data[i] = first + i;
1708     qsort (old_data, old_n, sizeof(int*), icvCmpIntegersPtr);
1709 
1710 // Count the number of different responses.
1711     for (i = 0, j = 0, out_n = 0; i < old_n && j < new_n; out_n++)
1712     {
1713         if (*old_data[i] == *new_data[j])
1714         {
1715             i++;
1716             j++;
1717         }
1718         else if (*old_data[i] < *new_data[j])
1719             i++;
1720         else
1721             j++;
1722     }
1723     out_n += old_n - i + new_n - j;
1724 
1725 // Create and fill the result response maps.
1726     CV_CALL (*out_response_map = cvCreateMat (1, out_n, CV_32SC1));
1727     out_data = (*out_response_map)->data.i;
1728     memcpy (out_data, first, old_n * sizeof (int));
1729 
1730     free_response = old_n;
1731     for (i = 0, j = 0; i < old_n && j < new_n; )
1732     {
1733         if (*old_data[i] == *new_data[j])
1734         {
1735             *new_data[j] = (int)(old_data[i] - first);
1736             i++;
1737             j++;
1738         }
1739         else if (*old_data[i] < *new_data[j])
1740             i++;
1741         else
1742         {
1743             out_data[free_response] = *new_data[j];
1744             *new_data[j] = free_response++;
1745             j++;
1746         }
1747     }
1748     for (; j < new_n; j++)
1749     {
1750         out_data[free_response] = *new_data[j];
1751         *new_data[j] = free_response++;
1752     }
1753     CV_ASSERT (free_response == out_n);
1754 
1755 // Change <responses> according to out response map.
1756     samples = _responses->cols + _responses->rows - 1;
1757     responses = _responses->data.i;
1758     first = new_response_map->data.i;
1759     for (i = 0; i < samples; i++)
1760     {
1761         responses[i] = first[responses[i]];
1762     }
1763 
1764         __END__
1765 
1766     cvFree(&old_data);
1767     cvFree(&new_data);
1768 
1769 }
1770 
1771 
icvGetNumberOfCluster(double * prob_vector,int num_of_clusters,float r,float outlier_thresh,int normalize_probs)1772 static int icvGetNumberOfCluster( double* prob_vector, int num_of_clusters, float r,
1773                            float outlier_thresh, int normalize_probs )
1774 {
1775     int max_prob_loc = 0;
1776 
1777     CV_FUNCNAME("icvGetNumberOfCluster");
1778     __BEGIN__;
1779 
1780     double prob, maxprob, sum;
1781     int i;
1782 
1783     CV_ASSERT(prob_vector);
1784     CV_ASSERT(num_of_clusters >= 0);
1785 
1786     maxprob = prob_vector[0];
1787     max_prob_loc = 0;
1788     sum = maxprob;
1789     for( i = 1; i < num_of_clusters; i++ )
1790     {
1791         prob = prob_vector[i];
1792         sum += prob;
1793         if( prob > maxprob )
1794         {
1795             max_prob_loc = i;
1796             maxprob = prob;
1797         }
1798     }
1799     if( normalize_probs && fabs(sum - 1.) > FLT_EPSILON )
1800     {
1801         for( i = 0; i < num_of_clusters; i++ )
1802             prob_vector[i] /= sum;
1803     }
1804     if( fabs(r - 1.) > FLT_EPSILON && fabs(sum - 1.) < outlier_thresh )
1805         max_prob_loc = -1;
1806 
1807     __END__;
1808 
1809     return max_prob_loc;
1810 
1811 } // End of icvGetNumberOfCluster
1812 
1813 
icvFindClusterLabels(const CvMat * probs,float outlier_thresh,float r,const CvMat * labels)1814 void icvFindClusterLabels( const CvMat* probs, float outlier_thresh, float r,
1815                           const CvMat* labels )
1816 {
1817     CvMat* counts = 0;
1818 
1819     CV_FUNCNAME("icvFindClusterLabels");
1820     __BEGIN__;
1821 
1822     int nclusters, nsamples;
1823     int i, j;
1824     double* probs_data;
1825 
1826     CV_ASSERT( ICV_IS_MAT_OF_TYPE(probs, CV_64FC1) );
1827     CV_ASSERT( ICV_IS_MAT_OF_TYPE(labels, CV_32SC1) );
1828 
1829     nclusters = probs->cols;
1830     nsamples  = probs->rows;
1831     CV_ASSERT( nsamples == labels->cols );
1832 
1833     CV_CALL( counts = cvCreateMat( 1, nclusters + 1, CV_32SC1 ) );
1834     CV_CALL( cvSetZero( counts ));
1835     for( i = 0; i < nsamples; i++ )
1836     {
1837         labels->data.i[i] = icvGetNumberOfCluster( probs->data.db + i*probs->cols,
1838             nclusters, r, outlier_thresh, 1 );
1839         counts->data.i[labels->data.i[i] + 1]++;
1840     }
1841     CV_ASSERT((int)cvSum(counts).val[0] == nsamples);
1842     // Filling empty clusters with the vector, that has the maximal probability
1843     for( j = 0; j < nclusters; j++ ) // outliers are ignored
1844     {
1845         int maxprob_loc = -1;
1846         double maxprob = 0;
1847 
1848         if( counts->data.i[j+1] ) // j-th class is not empty
1849             continue;
1850         // look for the presentative, which is not lonely in it's cluster
1851         // and that has a maximal probability among all these vectors
1852         probs_data = probs->data.db;
1853         for( i = 0; i < nsamples; i++, probs_data++ )
1854         {
1855             int label = labels->data.i[i];
1856             double prob;
1857             if( counts->data.i[label+1] == 0 ||
1858                 (counts->data.i[label+1] <= 1 && label != -1) )
1859                 continue;
1860             prob = *probs_data;
1861             if( prob >= maxprob )
1862             {
1863                 maxprob = prob;
1864                 maxprob_loc = i;
1865             }
1866         }
1867         // maxprob_loc == 0 <=> number of vectors less then number of clusters
1868         CV_ASSERT( maxprob_loc >= 0 );
1869         counts->data.i[labels->data.i[maxprob_loc] + 1]--;
1870         labels->data.i[maxprob_loc] = j;
1871         counts->data.i[j + 1]++;
1872     }
1873 
1874     __END__;
1875 
1876     cvReleaseMat( &counts );
1877 } // End of icvFindClusterLabels
1878 
1879 /* End of file */
1880