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 std;
45 using namespace cv;
46
47 const string FEATURES2D_DIR = "features2d";
48 const string IMAGE_FILENAME = "tsukuba.png";
49 const string DESCRIPTOR_DIR = FEATURES2D_DIR + "/descriptor_extractors";
50
51 /****************************************************************************************\
52 * Regression tests for descriptor extractors. *
53 \****************************************************************************************/
writeMatInBin(const Mat & mat,const string & filename)54 static void writeMatInBin( const Mat& mat, const string& filename )
55 {
56 FILE* f = fopen( filename.c_str(), "wb");
57 if( f )
58 {
59 int type = mat.type();
60 fwrite( (void*)&mat.rows, sizeof(int), 1, f );
61 fwrite( (void*)&mat.cols, sizeof(int), 1, f );
62 fwrite( (void*)&type, sizeof(int), 1, f );
63 int dataSize = (int)(mat.step * mat.rows * mat.channels());
64 fwrite( (void*)&dataSize, sizeof(int), 1, f );
65 fwrite( (void*)mat.ptr(), 1, dataSize, f );
66 fclose(f);
67 }
68 }
69
readMatFromBin(const string & filename)70 static Mat readMatFromBin( const string& filename )
71 {
72 FILE* f = fopen( filename.c_str(), "rb" );
73 if( f )
74 {
75 int rows, cols, type, dataSize;
76 size_t elements_read1 = fread( (void*)&rows, sizeof(int), 1, f );
77 size_t elements_read2 = fread( (void*)&cols, sizeof(int), 1, f );
78 size_t elements_read3 = fread( (void*)&type, sizeof(int), 1, f );
79 size_t elements_read4 = fread( (void*)&dataSize, sizeof(int), 1, f );
80 CV_Assert(elements_read1 == 1 && elements_read2 == 1 && elements_read3 == 1 && elements_read4 == 1);
81
82 int step = dataSize / rows / CV_ELEM_SIZE(type);
83 CV_Assert(step >= cols);
84
85 Mat m = Mat(rows, step, type).colRange(0, cols);
86
87 size_t elements_read = fread( m.ptr(), 1, dataSize, f );
88 CV_Assert(elements_read == (size_t)(dataSize));
89 fclose(f);
90
91 return m;
92 }
93 return Mat();
94 }
95
96 template<class Distance>
97 class CV_DescriptorExtractorTest : public cvtest::BaseTest
98 {
99 public:
100 typedef typename Distance::ValueType ValueType;
101 typedef typename Distance::ResultType DistanceType;
102
CV_DescriptorExtractorTest(const string _name,DistanceType _maxDist,const Ptr<DescriptorExtractor> & _dextractor,Distance d=Distance (),Ptr<FeatureDetector> _detector=Ptr<FeatureDetector> ())103 CV_DescriptorExtractorTest( const string _name, DistanceType _maxDist, const Ptr<DescriptorExtractor>& _dextractor,
104 Distance d = Distance(), Ptr<FeatureDetector> _detector = Ptr<FeatureDetector>()):
105 name(_name), maxDist(_maxDist), dextractor(_dextractor), distance(d) , detector(_detector) {}
106
~CV_DescriptorExtractorTest()107 ~CV_DescriptorExtractorTest()
108 {
109 }
110 protected:
createDescriptorExtractor()111 virtual void createDescriptorExtractor() {}
112
compareDescriptors(const Mat & validDescriptors,const Mat & calcDescriptors)113 void compareDescriptors( const Mat& validDescriptors, const Mat& calcDescriptors )
114 {
115 if( validDescriptors.size != calcDescriptors.size || validDescriptors.type() != calcDescriptors.type() )
116 {
117 ts->printf(cvtest::TS::LOG, "Valid and computed descriptors matrices must have the same size and type.\n");
118 ts->set_failed_test_info( cvtest::TS::FAIL_INVALID_TEST_DATA );
119 return;
120 }
121
122 CV_Assert( DataType<ValueType>::type == validDescriptors.type() );
123
124 int dimension = validDescriptors.cols;
125 DistanceType curMaxDist = std::numeric_limits<DistanceType>::min();
126 for( int y = 0; y < validDescriptors.rows; y++ )
127 {
128 DistanceType dist = distance( validDescriptors.ptr<ValueType>(y), calcDescriptors.ptr<ValueType>(y), dimension );
129 if( dist > curMaxDist )
130 curMaxDist = dist;
131 }
132
133 stringstream ss;
134 ss << "Max distance between valid and computed descriptors " << curMaxDist;
135 if( curMaxDist <= maxDist )
136 ss << "." << endl;
137 else
138 {
139 ss << ">" << maxDist << " - bad accuracy!"<< endl;
140 ts->set_failed_test_info( cvtest::TS::FAIL_BAD_ACCURACY );
141 }
142 ts->printf(cvtest::TS::LOG, ss.str().c_str() );
143 }
144
emptyDataTest()145 void emptyDataTest()
146 {
147 assert( dextractor );
148
149 // One image.
150 Mat image;
151 vector<KeyPoint> keypoints;
152 Mat descriptors;
153
154 try
155 {
156 dextractor->compute( image, keypoints, descriptors );
157 }
158 catch(...)
159 {
160 ts->printf( cvtest::TS::LOG, "compute() on empty image and empty keypoints must not generate exception (1).\n");
161 ts->set_failed_test_info( cvtest::TS::FAIL_INVALID_TEST_DATA );
162 }
163
164 image.create( 50, 50, CV_8UC3 );
165 try
166 {
167 dextractor->compute( image, keypoints, descriptors );
168 }
169 catch(...)
170 {
171 ts->printf( cvtest::TS::LOG, "compute() on nonempty image and empty keypoints must not generate exception (1).\n");
172 ts->set_failed_test_info( cvtest::TS::FAIL_INVALID_TEST_DATA );
173 }
174
175 // Several images.
176 vector<Mat> images;
177 vector<vector<KeyPoint> > keypointsCollection;
178 vector<Mat> descriptorsCollection;
179 try
180 {
181 dextractor->compute( images, keypointsCollection, descriptorsCollection );
182 }
183 catch(...)
184 {
185 ts->printf( cvtest::TS::LOG, "compute() on empty images and empty keypoints collection must not generate exception (2).\n");
186 ts->set_failed_test_info( cvtest::TS::FAIL_INVALID_TEST_DATA );
187 }
188 }
189
regressionTest()190 void regressionTest()
191 {
192 assert( dextractor );
193
194 // Read the test image.
195 string imgFilename = string(ts->get_data_path()) + FEATURES2D_DIR + "/" + IMAGE_FILENAME;
196 Mat img = imread( imgFilename );
197 if( img.empty() )
198 {
199 ts->printf( cvtest::TS::LOG, "Image %s can not be read.\n", imgFilename.c_str() );
200 ts->set_failed_test_info( cvtest::TS::FAIL_INVALID_TEST_DATA );
201 return;
202 }
203 vector<KeyPoint> keypoints;
204 FileStorage fs( string(ts->get_data_path()) + FEATURES2D_DIR + "/keypoints.xml.gz", FileStorage::READ );
205 if(!detector.empty()) {
206 detector->detect(img, keypoints);
207 } else {
208 read( fs.getFirstTopLevelNode(), keypoints );
209 }
210 if(!keypoints.empty())
211 {
212 Mat calcDescriptors;
213 double t = (double)getTickCount();
214 dextractor->compute( img, keypoints, calcDescriptors );
215 t = getTickCount() - t;
216 ts->printf(cvtest::TS::LOG, "\nAverage time of computing one descriptor = %g ms.\n", t/((double)getTickFrequency()*1000.)/calcDescriptors.rows);
217
218 if( calcDescriptors.rows != (int)keypoints.size() )
219 {
220 ts->printf( cvtest::TS::LOG, "Count of computed descriptors and keypoints count must be equal.\n" );
221 ts->printf( cvtest::TS::LOG, "Count of keypoints is %d.\n", (int)keypoints.size() );
222 ts->printf( cvtest::TS::LOG, "Count of computed descriptors is %d.\n", calcDescriptors.rows );
223 ts->set_failed_test_info( cvtest::TS::FAIL_INVALID_OUTPUT );
224 return;
225 }
226
227 if( calcDescriptors.cols != dextractor->descriptorSize() || calcDescriptors.type() != dextractor->descriptorType() )
228 {
229 ts->printf( cvtest::TS::LOG, "Incorrect descriptor size or descriptor type.\n" );
230 ts->printf( cvtest::TS::LOG, "Expected size is %d.\n", dextractor->descriptorSize() );
231 ts->printf( cvtest::TS::LOG, "Calculated size is %d.\n", calcDescriptors.cols );
232 ts->printf( cvtest::TS::LOG, "Expected type is %d.\n", dextractor->descriptorType() );
233 ts->printf( cvtest::TS::LOG, "Calculated type is %d.\n", calcDescriptors.type() );
234 ts->set_failed_test_info( cvtest::TS::FAIL_INVALID_OUTPUT );
235 return;
236 }
237
238 // TODO read and write descriptor extractor parameters and check them
239 Mat validDescriptors = readDescriptors();
240 if( !validDescriptors.empty() )
241 compareDescriptors( validDescriptors, calcDescriptors );
242 else
243 {
244 if( !writeDescriptors( calcDescriptors ) )
245 {
246 ts->printf( cvtest::TS::LOG, "Descriptors can not be written.\n" );
247 ts->set_failed_test_info( cvtest::TS::FAIL_INVALID_TEST_DATA );
248 return;
249 }
250 }
251 }
252 if(!fs.isOpened())
253 {
254 ts->printf( cvtest::TS::LOG, "Compute and write keypoints.\n" );
255 fs.open( string(ts->get_data_path()) + FEATURES2D_DIR + "/keypoints.xml.gz", FileStorage::WRITE );
256 if( fs.isOpened() )
257 {
258 Ptr<ORB> fd = ORB::create();
259 fd->detect(img, keypoints);
260 write( fs, "keypoints", keypoints );
261 }
262 else
263 {
264 ts->printf(cvtest::TS::LOG, "File for writting keypoints can not be opened.\n");
265 ts->set_failed_test_info( cvtest::TS::FAIL_INVALID_TEST_DATA );
266 return;
267 }
268 }
269 }
270
run(int)271 void run(int)
272 {
273 createDescriptorExtractor();
274 if( !dextractor )
275 {
276 ts->printf(cvtest::TS::LOG, "Descriptor extractor is empty.\n");
277 ts->set_failed_test_info( cvtest::TS::FAIL_INVALID_TEST_DATA );
278 return;
279 }
280
281 emptyDataTest();
282 regressionTest();
283
284 ts->set_failed_test_info( cvtest::TS::OK );
285 }
286
readDescriptors()287 virtual Mat readDescriptors()
288 {
289 Mat res = readMatFromBin( string(ts->get_data_path()) + DESCRIPTOR_DIR + "/" + string(name) );
290 return res;
291 }
292
writeDescriptors(Mat & descs)293 virtual bool writeDescriptors( Mat& descs )
294 {
295 writeMatInBin( descs, string(ts->get_data_path()) + DESCRIPTOR_DIR + "/" + string(name) );
296 return true;
297 }
298
299 string name;
300 const DistanceType maxDist;
301 Ptr<DescriptorExtractor> dextractor;
302 Distance distance;
303 Ptr<FeatureDetector> detector;
304
305 private:
operator =(const CV_DescriptorExtractorTest &)306 CV_DescriptorExtractorTest& operator=(const CV_DescriptorExtractorTest&) { return *this; }
307 };
308
309 /****************************************************************************************\
310 * Tests registrations *
311 \****************************************************************************************/
312
TEST(Features2d_DescriptorExtractor_BRISK,regression)313 TEST( Features2d_DescriptorExtractor_BRISK, regression )
314 {
315 CV_DescriptorExtractorTest<Hamming> test( "descriptor-brisk",
316 (CV_DescriptorExtractorTest<Hamming>::DistanceType)2.f,
317 BRISK::create() );
318 test.safe_run();
319 }
320
TEST(Features2d_DescriptorExtractor_ORB,regression)321 TEST( Features2d_DescriptorExtractor_ORB, regression )
322 {
323 // TODO adjust the parameters below
324 CV_DescriptorExtractorTest<Hamming> test( "descriptor-orb",
325 #if CV_NEON
326 (CV_DescriptorExtractorTest<Hamming>::DistanceType)25.f,
327 #else
328 (CV_DescriptorExtractorTest<Hamming>::DistanceType)12.f,
329 #endif
330 ORB::create() );
331 test.safe_run();
332 }
333
TEST(Features2d_DescriptorExtractor_KAZE,regression)334 TEST( Features2d_DescriptorExtractor_KAZE, regression )
335 {
336 CV_DescriptorExtractorTest< L2<float> > test( "descriptor-kaze", 0.03f,
337 KAZE::create(),
338 L2<float>(), KAZE::create() );
339 test.safe_run();
340 }
341
TEST(Features2d_DescriptorExtractor_AKAZE,regression)342 TEST( Features2d_DescriptorExtractor_AKAZE, regression )
343 {
344 CV_DescriptorExtractorTest<Hamming> test( "descriptor-akaze",
345 (CV_DescriptorExtractorTest<Hamming>::DistanceType)12.f,
346 AKAZE::create(),
347 Hamming(), AKAZE::create());
348 test.safe_run();
349 }
350
TEST(Features2d_DescriptorExtractor,batch)351 TEST( Features2d_DescriptorExtractor, batch )
352 {
353 string path = string(cvtest::TS::ptr()->get_data_path() + "detectors_descriptors_evaluation/images_datasets/graf");
354 vector<Mat> imgs, descriptors;
355 vector<vector<KeyPoint> > keypoints;
356 int i, n = 6;
357 Ptr<ORB> orb = ORB::create();
358
359 for( i = 0; i < n; i++ )
360 {
361 string imgname = format("%s/img%d.png", path.c_str(), i+1);
362 Mat img = imread(imgname, 0);
363 imgs.push_back(img);
364 }
365
366 orb->detect(imgs, keypoints);
367 orb->compute(imgs, keypoints, descriptors);
368
369 ASSERT_EQ((int)keypoints.size(), n);
370 ASSERT_EQ((int)descriptors.size(), n);
371
372 for( i = 0; i < n; i++ )
373 {
374 EXPECT_GT((int)keypoints[i].size(), 100);
375 EXPECT_GT(descriptors[i].rows, 100);
376 }
377 }
378
TEST(Features2d_Feature2d,no_crash)379 TEST( Features2d_Feature2d, no_crash )
380 {
381 const String& pattern = string(cvtest::TS::ptr()->get_data_path() + "shared/*.png");
382 vector<String> fnames;
383 glob(pattern, fnames, false);
384 sort(fnames.begin(), fnames.end());
385
386 Ptr<AKAZE> akaze = AKAZE::create();
387 Ptr<ORB> orb = ORB::create();
388 Ptr<KAZE> kaze = KAZE::create();
389 Ptr<BRISK> brisk = BRISK::create();
390 size_t i, n = fnames.size();
391 vector<KeyPoint> keypoints;
392 Mat descriptors;
393 orb->setMaxFeatures(5000);
394
395 for( i = 0; i < n; i++ )
396 {
397 printf("%d. image: %s:\n", (int)i, fnames[i].c_str());
398 if( strstr(fnames[i].c_str(), "MP.png") != 0 )
399 continue;
400 bool checkCount = strstr(fnames[i].c_str(), "templ.png") == 0;
401
402 Mat img = imread(fnames[i], -1);
403 printf("\tAKAZE ... "); fflush(stdout);
404 akaze->detectAndCompute(img, noArray(), keypoints, descriptors);
405 printf("(%d keypoints) ", (int)keypoints.size()); fflush(stdout);
406 if( checkCount )
407 {
408 EXPECT_GT((int)keypoints.size(), 0);
409 }
410 ASSERT_EQ(descriptors.rows, (int)keypoints.size());
411 printf("ok\n");
412
413 printf("\tKAZE ... "); fflush(stdout);
414 kaze->detectAndCompute(img, noArray(), keypoints, descriptors);
415 printf("(%d keypoints) ", (int)keypoints.size()); fflush(stdout);
416 if( checkCount )
417 {
418 EXPECT_GT((int)keypoints.size(), 0);
419 }
420 ASSERT_EQ(descriptors.rows, (int)keypoints.size());
421 printf("ok\n");
422
423 printf("\tORB ... "); fflush(stdout);
424 orb->detectAndCompute(img, noArray(), keypoints, descriptors);
425 printf("(%d keypoints) ", (int)keypoints.size()); fflush(stdout);
426 if( checkCount )
427 {
428 EXPECT_GT((int)keypoints.size(), 0);
429 }
430 ASSERT_EQ(descriptors.rows, (int)keypoints.size());
431 printf("ok\n");
432
433 printf("\tBRISK ... "); fflush(stdout);
434 brisk->detectAndCompute(img, noArray(), keypoints, descriptors);
435 printf("(%d keypoints) ", (int)keypoints.size()); fflush(stdout);
436 if( checkCount )
437 {
438 EXPECT_GT((int)keypoints.size(), 0);
439 }
440 ASSERT_EQ(descriptors.rows, (int)keypoints.size());
441 printf("ok\n");
442 }
443 }
444