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 //                For Open Source Computer Vision Library
12 //
13 // Copyright (C) 2000, Intel Corporation, all rights reserved.
14 // Third party copyrights are property of their respective owners.
15 //
16 // Redistribution and use in source and binary forms, with or without modification,
17 // are permitted provided that the following conditions are met:
18 //
19 //   * Redistribution's of source code must retain the above copyright notice,
20 //     this list of conditions and the following disclaimer.
21 //
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.
25 //
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.
28 //
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.
39 //
40 //M*/
41 
42 #include "test_precomp.hpp"
43 
44 using namespace cv;
45 using namespace std;
46 
47 class CV_TemplMatchTest : public cvtest::ArrayTest
48 {
49 public:
50     CV_TemplMatchTest();
51 
52 protected:
53     int read_params( CvFileStorage* fs );
54     void get_test_array_types_and_sizes( int test_case_idx, vector<vector<Size> >& sizes, vector<vector<int> >& types );
55     void get_minmax_bounds( int i, int j, int type, Scalar& low, Scalar& high );
56     double get_success_error_level( int test_case_idx, int i, int j );
57     void run_func();
58     void prepare_to_validation( int );
59 
60     int max_template_size;
61     int method;
62     bool test_cpp;
63 };
64 
65 
CV_TemplMatchTest()66 CV_TemplMatchTest::CV_TemplMatchTest()
67 {
68     test_array[INPUT].push_back(NULL);
69     test_array[INPUT].push_back(NULL);
70     test_array[OUTPUT].push_back(NULL);
71     test_array[REF_OUTPUT].push_back(NULL);
72     element_wise_relative_error = false;
73     max_template_size = 100;
74     method = 0;
75     test_cpp = false;
76 }
77 
78 
read_params(CvFileStorage * fs)79 int CV_TemplMatchTest::read_params( CvFileStorage* fs )
80 {
81     int code = cvtest::ArrayTest::read_params( fs );
82     if( code < 0 )
83         return code;
84 
85     max_template_size = cvReadInt( find_param( fs, "max_template_size" ), max_template_size );
86     max_template_size = cvtest::clipInt( max_template_size, 1, 100 );
87 
88     return code;
89 }
90 
91 
get_minmax_bounds(int i,int j,int type,Scalar & low,Scalar & high)92 void CV_TemplMatchTest::get_minmax_bounds( int i, int j, int type, Scalar& low, Scalar& high )
93 {
94     cvtest::ArrayTest::get_minmax_bounds( i, j, type, low, high );
95     int depth = CV_MAT_DEPTH(type);
96     if( depth == CV_32F )
97     {
98         low = Scalar::all(-10.);
99         high = Scalar::all(10.);
100     }
101 }
102 
103 
get_test_array_types_and_sizes(int test_case_idx,vector<vector<Size>> & sizes,vector<vector<int>> & types)104 void CV_TemplMatchTest::get_test_array_types_and_sizes( int test_case_idx,
105                                                 vector<vector<Size> >& sizes, vector<vector<int> >& types )
106 {
107     RNG& rng = ts->get_rng();
108     int depth = cvtest::randInt(rng) % 2, cn = cvtest::randInt(rng) & 1 ? 3 : 1;
109     cvtest::ArrayTest::get_test_array_types_and_sizes( test_case_idx, sizes, types );
110     depth = depth == 0 ? CV_8U : CV_32F;
111 
112     types[INPUT][0] = types[INPUT][1] = CV_MAKETYPE(depth,cn);
113     types[OUTPUT][0] = types[REF_OUTPUT][0] = CV_32FC1;
114 
115     sizes[INPUT][1].width = cvtest::randInt(rng)%MIN(sizes[INPUT][1].width,max_template_size) + 1;
116     sizes[INPUT][1].height = cvtest::randInt(rng)%MIN(sizes[INPUT][1].height,max_template_size) + 1;
117     sizes[OUTPUT][0].width = sizes[INPUT][0].width - sizes[INPUT][1].width + 1;
118     sizes[OUTPUT][0].height = sizes[INPUT][0].height - sizes[INPUT][1].height + 1;
119     sizes[REF_OUTPUT][0] = sizes[OUTPUT][0];
120 
121     method = cvtest::randInt(rng)%6;
122     test_cpp = (cvtest::randInt(rng) & 256) == 0;
123 }
124 
125 
get_success_error_level(int,int,int)126 double CV_TemplMatchTest::get_success_error_level( int /*test_case_idx*/, int /*i*/, int /*j*/ )
127 {
128     if( test_mat[INPUT][1].depth() == CV_8U ||
129         (method >= CV_TM_CCOEFF && test_mat[INPUT][1].cols*test_mat[INPUT][1].rows <= 2) )
130         return 1e-2;
131     else
132         return 1e-3;
133 }
134 
135 
run_func()136 void CV_TemplMatchTest::run_func()
137 {
138     if(!test_cpp)
139         cvMatchTemplate( test_array[INPUT][0], test_array[INPUT][1], test_array[OUTPUT][0], method );
140     else
141     {
142         cv::Mat _out = cv::cvarrToMat(test_array[OUTPUT][0]);
143         cv::matchTemplate(cv::cvarrToMat(test_array[INPUT][0]), cv::cvarrToMat(test_array[INPUT][1]), _out, method);
144     }
145 }
146 
147 
cvTsMatchTemplate(const CvMat * img,const CvMat * templ,CvMat * result,int method)148 static void cvTsMatchTemplate( const CvMat* img, const CvMat* templ, CvMat* result, int method )
149 {
150     int i, j, k, l;
151     int depth = CV_MAT_DEPTH(img->type), cn = CV_MAT_CN(img->type);
152     int width_n = templ->cols*cn, height = templ->rows;
153     int a_step = img->step / CV_ELEM_SIZE(img->type & CV_MAT_DEPTH_MASK);
154     int b_step = templ->step / CV_ELEM_SIZE(templ->type & CV_MAT_DEPTH_MASK);
155     CvScalar b_mean, b_sdv;
156     double b_denom = 1., b_sum2 = 0;
157     int area = templ->rows*templ->cols;
158 
159     cvAvgSdv(templ, &b_mean, &b_sdv);
160 
161     for( i = 0; i < cn; i++ )
162         b_sum2 += (b_sdv.val[i]*b_sdv.val[i] + b_mean.val[i]*b_mean.val[i])*area;
163 
164     if( b_sdv.val[0]*b_sdv.val[0] + b_sdv.val[1]*b_sdv.val[1] +
165         b_sdv.val[2]*b_sdv.val[2] + b_sdv.val[3]*b_sdv.val[3] < DBL_EPSILON &&
166         method == CV_TM_CCOEFF_NORMED )
167     {
168         cvSet( result, cvScalarAll(1.) );
169         return;
170     }
171 
172     if( method & 1 )
173     {
174         b_denom = 0;
175         if( method != CV_TM_CCOEFF_NORMED )
176         {
177             b_denom = b_sum2;
178         }
179         else
180         {
181             for( i = 0; i < cn; i++ )
182                 b_denom += b_sdv.val[i]*b_sdv.val[i]*area;
183         }
184         b_denom = sqrt(b_denom);
185         if( b_denom == 0 )
186             b_denom = 1.;
187     }
188 
189     assert( CV_TM_SQDIFF <= method && method <= CV_TM_CCOEFF_NORMED );
190 
191     for( i = 0; i < result->rows; i++ )
192     {
193         for( j = 0; j < result->cols; j++ )
194         {
195             CvScalar a_sum(0), a_sum2(0);
196             CvScalar ccorr(0);
197             double value = 0.;
198 
199             if( depth == CV_8U )
200             {
201                 const uchar* a = img->data.ptr + i*img->step + j*cn;
202                 const uchar* b = templ->data.ptr;
203 
204                 if( cn == 1 || method < CV_TM_CCOEFF )
205                 {
206                     for( k = 0; k < height; k++, a += a_step, b += b_step )
207                         for( l = 0; l < width_n; l++ )
208                         {
209                             ccorr.val[0] += a[l]*b[l];
210                             a_sum.val[0] += a[l];
211                             a_sum2.val[0] += a[l]*a[l];
212                         }
213                 }
214                 else
215                 {
216                     for( k = 0; k < height; k++, a += a_step, b += b_step )
217                         for( l = 0; l < width_n; l += 3 )
218                         {
219                             ccorr.val[0] += a[l]*b[l];
220                             ccorr.val[1] += a[l+1]*b[l+1];
221                             ccorr.val[2] += a[l+2]*b[l+2];
222                             a_sum.val[0] += a[l];
223                             a_sum.val[1] += a[l+1];
224                             a_sum.val[2] += a[l+2];
225                             a_sum2.val[0] += a[l]*a[l];
226                             a_sum2.val[1] += a[l+1]*a[l+1];
227                             a_sum2.val[2] += a[l+2]*a[l+2];
228                         }
229                 }
230             }
231             else
232             {
233                 const float* a = (const float*)(img->data.ptr + i*img->step) + j*cn;
234                 const float* b = (const float*)templ->data.ptr;
235 
236                 if( cn == 1 || method < CV_TM_CCOEFF )
237                 {
238                     for( k = 0; k < height; k++, a += a_step, b += b_step )
239                         for( l = 0; l < width_n; l++ )
240                         {
241                             ccorr.val[0] += a[l]*b[l];
242                             a_sum.val[0] += a[l];
243                             a_sum2.val[0] += a[l]*a[l];
244                         }
245                 }
246                 else
247                 {
248                     for( k = 0; k < height; k++, a += a_step, b += b_step )
249                         for( l = 0; l < width_n; l += 3 )
250                         {
251                             ccorr.val[0] += a[l]*b[l];
252                             ccorr.val[1] += a[l+1]*b[l+1];
253                             ccorr.val[2] += a[l+2]*b[l+2];
254                             a_sum.val[0] += a[l];
255                             a_sum.val[1] += a[l+1];
256                             a_sum.val[2] += a[l+2];
257                             a_sum2.val[0] += a[l]*a[l];
258                             a_sum2.val[1] += a[l+1]*a[l+1];
259                             a_sum2.val[2] += a[l+2]*a[l+2];
260                         }
261                 }
262             }
263 
264             switch( method )
265             {
266             case CV_TM_CCORR:
267             case CV_TM_CCORR_NORMED:
268                 value = ccorr.val[0];
269                 break;
270             case CV_TM_SQDIFF:
271             case CV_TM_SQDIFF_NORMED:
272                 value = (a_sum2.val[0] + b_sum2 - 2*ccorr.val[0]);
273                 break;
274             default:
275                 value = (ccorr.val[0] - a_sum.val[0]*b_mean.val[0]+
276                          ccorr.val[1] - a_sum.val[1]*b_mean.val[1]+
277                          ccorr.val[2] - a_sum.val[2]*b_mean.val[2]);
278             }
279 
280             if( method & 1 )
281             {
282                 double denom;
283 
284                 // calc denominator
285                 if( method != CV_TM_CCOEFF_NORMED )
286                 {
287                     denom = a_sum2.val[0] + a_sum2.val[1] + a_sum2.val[2];
288                 }
289                 else
290                 {
291                     denom = a_sum2.val[0] - (a_sum.val[0]*a_sum.val[0])/area;
292                     denom += a_sum2.val[1] - (a_sum.val[1]*a_sum.val[1])/area;
293                     denom += a_sum2.val[2] - (a_sum.val[2]*a_sum.val[2])/area;
294                 }
295                 denom = sqrt(MAX(denom,0))*b_denom;
296                 if( fabs(value) < denom )
297                     value /= denom;
298                 else if( fabs(value) < denom*1.125 )
299                     value = value > 0 ? 1 : -1;
300                 else
301                     value = method != CV_TM_SQDIFF_NORMED ? 0 : 1;
302             }
303 
304             ((float*)(result->data.ptr + result->step*i))[j] = (float)value;
305         }
306     }
307 }
308 
309 
prepare_to_validation(int)310 void CV_TemplMatchTest::prepare_to_validation( int /*test_case_idx*/ )
311 {
312     CvMat _input = test_mat[INPUT][0], _templ = test_mat[INPUT][1];
313     CvMat _output = test_mat[REF_OUTPUT][0];
314     cvTsMatchTemplate( &_input, &_templ, &_output, method );
315 
316     //if( ts->get_current_test_info()->test_case_idx == 0 )
317     /*{
318         CvFileStorage* fs = cvOpenFileStorage( "_match_template.yml", 0, CV_STORAGE_WRITE );
319         cvWrite( fs, "image", &test_mat[INPUT][0] );
320         cvWrite( fs, "template", &test_mat[INPUT][1] );
321         cvWrite( fs, "ref", &test_mat[REF_OUTPUT][0] );
322         cvWrite( fs, "opencv", &test_mat[OUTPUT][0] );
323         cvWriteInt( fs, "method", method );
324         cvReleaseFileStorage( &fs );
325     }*/
326 
327     if( method >= CV_TM_CCOEFF )
328     {
329         // avoid numerical stability problems in singular cases (when the results are near to 0)
330         const double delta = 10.;
331         test_mat[REF_OUTPUT][0] += Scalar::all(delta);
332         test_mat[OUTPUT][0] += Scalar::all(delta);
333     }
334 }
335 
TEST(Imgproc_MatchTemplate,accuracy)336 TEST(Imgproc_MatchTemplate, accuracy) { CV_TemplMatchTest test; test.safe_run(); }
337