Point Cloud Library (PCL)  1.14.1
color_gradient_modality.h
1 /*
2  * Software License Agreement (BSD License)
3  *
4  * Point Cloud Library (PCL) - www.pointclouds.org
5  * Copyright (c) 2010-2011, Willow Garage, Inc.
6  *
7  * All rights reserved.
8  *
9  * Redistribution and use in source and binary forms, with or without
10  * modification, are permitted provided that the following conditions
11  * are met:
12  *
13  * * Redistributions of source code must retain the above copyright
14  * notice, this list of conditions and the following disclaimer.
15  * * Redistributions in binary form must reproduce the above
16  * copyright notice, this list of conditions and the following
17  * disclaimer in the documentation and/or other materials provided
18  * with the distribution.
19  * * Neither the name of Willow Garage, Inc. nor the names of its
20  * contributors may be used to endorse or promote products derived
21  * from this software without specific prior written permission.
22  *
23  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
24  * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
25  * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
26  * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
27  * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
28  * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
29  * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
30  * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
31  * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
32  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
33  * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
34  * POSSIBILITY OF SUCH DAMAGE.
35  *
36  */
37 
38 #pragma once
39 
40 #include <pcl/recognition/quantizable_modality.h>
41 
42 #include <pcl/pcl_base.h>
43 #include <pcl/point_cloud.h>
44 #include <pcl/point_types.h>
45 #include <pcl/recognition/point_types.h>
46 #include <pcl/filters/convolution.h>
47 
48 #include <list>
49 
50 namespace pcl
51 {
52 
53  /** \brief Modality based on max-RGB gradients.
54  * \author Stefan Holzer
55  */
56  template <typename PointInT>
58  : public QuantizableModality, public PCLBase<PointInT>
59  {
60  protected:
62 
63  /** \brief Candidate for a feature (used in feature extraction methods). */
64  struct Candidate
65  {
66  /** \brief The gradient. */
68 
69  /** \brief The x-position. */
70  int x;
71  /** \brief The y-position. */
72  int y;
73 
74  /** \brief Operator for comparing to candidates (by magnitude of the gradient).
75  * \param[in] rhs the candidate to compare with.
76  */
77  bool operator< (const Candidate & rhs) const
78  {
79  return (gradient.magnitude > rhs.gradient.magnitude);
80  }
81  };
82 
83  public:
85 
86  /** \brief Different methods for feature selection/extraction. */
88  {
90  MASK_BORDER_EQUALLY, // this gives templates most equally to the OpenCV implementation
92  };
93 
94  /** \brief Constructor. */
96  /** \brief Destructor. */
97  ~ColorGradientModality () override;
98 
99  /** \brief Sets the threshold for the gradient magnitude which is used when quantizing the data.
100  * Gradients with a smaller magnitude are ignored.
101  * \param[in] threshold the new gradient magnitude threshold.
102  */
103  inline void
104  setGradientMagnitudeThreshold (const float threshold)
105  {
106  gradient_magnitude_threshold_ = threshold;
107  }
108 
109  /** \brief Sets the threshold for the gradient magnitude which is used for feature extraction.
110  * Gradients with a smaller magnitude are ignored.
111  * \param[in] threshold the new gradient magnitude threshold.
112  */
113  inline void
115  {
116  gradient_magnitude_threshold_feature_extraction_ = threshold;
117  }
118 
119  /** \brief Sets the feature selection method.
120  * \param[in] method the feature selection method.
121  */
122  inline void
124  {
125  feature_selection_method_ = method;
126  }
127 
128  /** \brief Sets the spreading size for spreading the quantized data. */
129  inline void
130  setSpreadingSize (const std::size_t spreading_size)
131  {
132  spreading_size_ = spreading_size;
133  }
134 
135  /** \brief Sets whether variable feature numbers for feature extraction is enabled.
136  * \param[in] enabled enables/disables variable feature numbers for feature extraction.
137  */
138  inline void
139  setVariableFeatureNr (const bool enabled)
140  {
141  variable_feature_nr_ = enabled;
142  }
143 
144  /** \brief Returns a reference to the internally computed quantized map. */
145  inline QuantizedMap &
146  getQuantizedMap () override
147  {
148  return (filtered_quantized_color_gradients_);
149  }
150 
151  /** \brief Returns a reference to the internally computed spread quantized map. */
152  inline QuantizedMap &
154  {
155  return (spreaded_filtered_quantized_color_gradients_);
156  }
157 
158  /** \brief Returns a point cloud containing the max-RGB gradients. */
161  {
162  return (color_gradients_);
163  }
164 
165  /** \brief Extracts features from this modality within the specified mask.
166  * \param[in] mask defines the areas where features are searched in.
167  * \param[in] nr_features defines the number of features to be extracted
168  * (might be less if not sufficient information is present in the modality).
169  * \param[in] modalityIndex the index which is stored in the extracted features.
170  * \param[out] features the destination for the extracted features.
171  */
172  void
173  extractFeatures (const MaskMap & mask, std::size_t nr_features, std::size_t modalityIndex,
174  std::vector<QuantizedMultiModFeature> & features) const override;
175 
176  /** \brief Extracts all possible features from the modality within the specified mask.
177  * \param[in] mask defines the areas where features are searched in.
178  * \param[in] nr_features IGNORED (TODO: remove this parameter).
179  * \param[in] modalityIndex the index which is stored in the extracted features.
180  * \param[out] features the destination for the extracted features.
181  */
182  void
183  extractAllFeatures (const MaskMap & mask, std::size_t nr_features, std::size_t modalityIndex,
184  std::vector<QuantizedMultiModFeature> & features) const override;
185 
186  /** \brief Provide a pointer to the input dataset (overwrites the PCLBase::setInputCloud method)
187  * \param cloud the const boost shared pointer to a PointCloud message
188  */
189  void
190  setInputCloud (const typename PointCloudIn::ConstPtr & cloud) override
191  {
192  input_ = cloud;
193  }
194 
195  /** \brief Processes the input data (smoothing, computing gradients, quantizing, filtering, spreading). */
196  virtual void
197  processInputData ();
198 
199  /** \brief Processes the input data assuming that everything up to filtering is already done/available
200  * (so only spreading is performed). */
201  virtual void
203 
204  protected:
205 
206  /** \brief Computes the Gaussian kernel used for smoothing.
207  * \param[in] kernel_size the size of the Gaussian kernel.
208  * \param[in] sigma the sigma.
209  * \param[out] kernel_values the destination for the values of the kernel. */
210  void
211  computeGaussianKernel (const std::size_t kernel_size, const float sigma, std::vector <float> & kernel_values);
212 
213  /** \brief Computes the max-RGB gradients for the specified cloud.
214  * \param[in] cloud the cloud for which the gradients are computed.
215  */
216  void
218 
219  /** \brief Computes the max-RGB gradients for the specified cloud using sobel.
220  * \param[in] cloud the cloud for which the gradients are computed.
221  */
222  void
224 
225  /** \brief Quantizes the color gradients. */
226  void
228 
229  /** \brief Filters the quantized gradients. */
230  void
232 
233  /** \brief Erodes a mask.
234  * \param[in] mask_in the mask which will be eroded.
235  * \param[out] mask_out the destination for the eroded mask.
236  */
237  static void
238  erode (const pcl::MaskMap & mask_in, pcl::MaskMap & mask_out);
239 
240  private:
241 
242  /** \brief Determines whether variable numbers of features are extracted or not. */
243  bool variable_feature_nr_{false};
244 
245  /** \brief Stores a smoothed version of the input cloud. */
246  pcl::PointCloud<pcl::RGB>::Ptr smoothed_input_;
247 
248  /** \brief Defines which feature selection method is used. */
249  FeatureSelectionMethod feature_selection_method_;
250 
251  /** \brief The threshold applied on the gradient magnitudes (for quantization). */
252  float gradient_magnitude_threshold_{10.0f};
253  /** \brief The threshold applied on the gradient magnitudes for feature extraction. */
254  float gradient_magnitude_threshold_feature_extraction_{55.0f};
255 
256  /** \brief The point cloud which holds the max-RGB gradients. */
257  pcl::PointCloud<pcl::GradientXY> color_gradients_;
258 
259  /** \brief The spreading size. */
260  std::size_t spreading_size_{8};
261 
262  /** \brief The map which holds the quantized max-RGB gradients. */
263  pcl::QuantizedMap quantized_color_gradients_;
264  /** \brief The map which holds the filtered quantized data. */
265  pcl::QuantizedMap filtered_quantized_color_gradients_;
266  /** \brief The map which holds the spread quantized data. */
267  pcl::QuantizedMap spreaded_filtered_quantized_color_gradients_;
268 
269  };
270 
271 }
272 
273 //////////////////////////////////////////////////////////////////////////////////////////////
274 template <typename PointInT>
277  : smoothed_input_ (new pcl::PointCloud<pcl::RGB> ())
278  , feature_selection_method_ (DISTANCE_MAGNITUDE_SCORE)
279 {
280 }
281 
282 //////////////////////////////////////////////////////////////////////////////////////////////
283 template <typename PointInT>
285 ~ColorGradientModality () = default;
286 
287 //////////////////////////////////////////////////////////////////////////////////////////////
288 template <typename PointInT> void
290 computeGaussianKernel (const std::size_t kernel_size, const float sigma, std::vector <float> & kernel_values)
291 {
292  // code taken from OpenCV
293  const int n = static_cast<int>(kernel_size);
294  constexpr int SMALL_GAUSSIAN_SIZE = 7;
295  static const float small_gaussian_tab[][SMALL_GAUSSIAN_SIZE] =
296  {
297  {1.f},
298  {0.25f, 0.5f, 0.25f},
299  {0.0625f, 0.25f, 0.375f, 0.25f, 0.0625f},
300  {0.03125f, 0.109375f, 0.21875f, 0.28125f, 0.21875f, 0.109375f, 0.03125f}
301  };
302 
303  const float* fixed_kernel = n % 2 == 1 && n <= SMALL_GAUSSIAN_SIZE && sigma <= 0 ?
304  small_gaussian_tab[n>>1] : nullptr;
305 
306  //CV_Assert( ktype == CV_32F || ktype == CV_64F );
307  /*Mat kernel(n, 1, ktype);*/
308  kernel_values.resize (n);
309  float* cf = kernel_values.data();
310  //double* cd = (double*)kernel.data;
311 
312  double sigmaX = sigma > 0 ? sigma : ((n-1)*0.5 - 1)*0.3 + 0.8;
313  double scale2X = -0.5/(sigmaX*sigmaX);
314  double sum = 0;
315 
316  for( int i = 0; i < n; i++ )
317  {
318  double x = i - (n-1)*0.5;
319  double t = fixed_kernel ? static_cast<double>(fixed_kernel[i]) : std::exp (scale2X*x*x);
320 
321  cf[i] = static_cast<float>(t);
322  sum += cf[i];
323  }
324 
325  sum = 1./sum;
326  for ( int i = 0; i < n; i++ )
327  {
328  cf[i] = static_cast<float>(cf[i]*sum);
329  }
330 }
331 
332 //////////////////////////////////////////////////////////////////////////////////////////////
333 template <typename PointInT>
334 void
337 {
338  // compute gaussian kernel values
339  constexpr std::size_t kernel_size = 7;
340  std::vector<float> kernel_values;
341  computeGaussianKernel (kernel_size, 0.0f, kernel_values);
342 
343  // smooth input
345  Eigen::ArrayXf gaussian_kernel(kernel_size);
346  //gaussian_kernel << 1.f/16, 1.f/8, 3.f/16, 2.f/8, 3.f/16, 1.f/8, 1.f/16;
347  //gaussian_kernel << 16.f/1600.f, 32.f/1600.f, 64.f/1600.f, 128.f/1600.f, 256.f/1600.f, 128.f/1600.f, 64.f/1600.f, 32.f/1600.f, 16.f/1600.f;
348  gaussian_kernel << kernel_values[0], kernel_values[1], kernel_values[2], kernel_values[3], kernel_values[4], kernel_values[5], kernel_values[6];
349 
351 
352  const std::uint32_t width = input_->width;
353  const std::uint32_t height = input_->height;
354 
355  rgb_input_->resize (width*height);
356  rgb_input_->width = width;
357  rgb_input_->height = height;
358  rgb_input_->is_dense = input_->is_dense;
359  for (std::size_t row_index = 0; row_index < height; ++row_index)
360  {
361  for (std::size_t col_index = 0; col_index < width; ++col_index)
362  {
363  (*rgb_input_) (col_index, row_index).r = (*input_) (col_index, row_index).r;
364  (*rgb_input_) (col_index, row_index).g = (*input_) (col_index, row_index).g;
365  (*rgb_input_) (col_index, row_index).b = (*input_) (col_index, row_index).b;
366  }
367  }
368 
369  convolution.setInputCloud (rgb_input_);
370  convolution.setKernel (gaussian_kernel);
371 
372  convolution.convolve (*smoothed_input_);
373 
374  // extract color gradients
375  computeMaxColorGradientsSobel (smoothed_input_);
376 
377  // quantize gradients
378  quantizeColorGradients ();
379 
380  // filter quantized gradients to get only dominants one + thresholding
381  filterQuantizedColorGradients ();
382 
383  // spread filtered quantized gradients
384  //spreadFilteredQunatizedColorGradients ();
385  pcl::QuantizedMap::spreadQuantizedMap (filtered_quantized_color_gradients_,
386  spreaded_filtered_quantized_color_gradients_,
387  spreading_size_);
388 }
389 
390 //////////////////////////////////////////////////////////////////////////////////////////////
391 template <typename PointInT>
392 void
395 {
396  // spread filtered quantized gradients
397  //spreadFilteredQunatizedColorGradients ();
398  pcl::QuantizedMap::spreadQuantizedMap (filtered_quantized_color_gradients_,
399  spreaded_filtered_quantized_color_gradients_,
400  spreading_size_);
401 }
402 
403 //////////////////////////////////////////////////////////////////////////////////////////////
404 template <typename PointInT>
406 extractFeatures (const MaskMap & mask, const std::size_t nr_features, const std::size_t modality_index,
407  std::vector<QuantizedMultiModFeature> & features) const
408 {
409  const std::size_t width = mask.getWidth ();
410  const std::size_t height = mask.getHeight ();
411 
412  std::list<Candidate> list1;
413  std::list<Candidate> list2;
414 
415 
416  if (feature_selection_method_ == DISTANCE_MAGNITUDE_SCORE)
417  {
418  for (std::size_t row_index = 0; row_index < height; ++row_index)
419  {
420  for (std::size_t col_index = 0; col_index < width; ++col_index)
421  {
422  if (mask (col_index, row_index) != 0)
423  {
424  const GradientXY & gradient = color_gradients_ (col_index, row_index);
425  if (gradient.magnitude > gradient_magnitude_threshold_feature_extraction_
426  && filtered_quantized_color_gradients_ (col_index, row_index) != 0)
427  {
428  Candidate candidate;
429  candidate.gradient = gradient;
430  candidate.x = static_cast<int> (col_index);
431  candidate.y = static_cast<int> (row_index);
432 
433  list1.push_back (candidate);
434  }
435  }
436  }
437  }
438 
439  list1.sort();
440 
441  if (variable_feature_nr_)
442  {
443  list2.push_back (*(list1.begin ()));
444  //while (list2.size () != nr_features)
445  bool feature_selection_finished = false;
446  while (!feature_selection_finished)
447  {
448  float best_score = 0.0f;
449  auto best_iter = list1.end ();
450  for (auto iter1 = list1.begin (); iter1 != list1.end (); ++iter1)
451  {
452  // find smallest distance
453  float smallest_distance = std::numeric_limits<float>::max ();
454  for (auto iter2 = list2.begin (); iter2 != list2.end (); ++iter2)
455  {
456  const float dx = static_cast<float> (iter1->x) - static_cast<float> (iter2->x);
457  const float dy = static_cast<float> (iter1->y) - static_cast<float> (iter2->y);
458 
459  const float distance = dx*dx + dy*dy;
460 
461  if (distance < smallest_distance)
462  {
463  smallest_distance = distance;
464  }
465  }
466 
467  const float score = smallest_distance * iter1->gradient.magnitude;
468 
469  if (score > best_score)
470  {
471  best_score = score;
472  best_iter = iter1;
473  }
474  }
475 
476 
477  float min_min_sqr_distance = std::numeric_limits<float>::max ();
478  float max_min_sqr_distance = 0;
479  for (auto iter2 = list2.begin (); iter2 != list2.end (); ++iter2)
480  {
481  float min_sqr_distance = std::numeric_limits<float>::max ();
482  for (auto iter3 = list2.begin (); iter3 != list2.end (); ++iter3)
483  {
484  if (iter2 == iter3)
485  continue;
486 
487  const float dx = static_cast<float> (iter2->x) - static_cast<float> (iter3->x);
488  const float dy = static_cast<float> (iter2->y) - static_cast<float> (iter3->y);
489 
490  const float sqr_distance = dx*dx + dy*dy;
491 
492  if (sqr_distance < min_sqr_distance)
493  {
494  min_sqr_distance = sqr_distance;
495  }
496 
497  //std::cerr << min_sqr_distance;
498  }
499  //std::cerr << std::endl;
500 
501  // check current feature
502  {
503  const float dx = static_cast<float> (iter2->x) - static_cast<float> (best_iter->x);
504  const float dy = static_cast<float> (iter2->y) - static_cast<float> (best_iter->y);
505 
506  const float sqr_distance = dx*dx + dy*dy;
507 
508  if (sqr_distance < min_sqr_distance)
509  {
510  min_sqr_distance = sqr_distance;
511  }
512  }
513 
514  if (min_sqr_distance < min_min_sqr_distance)
515  min_min_sqr_distance = min_sqr_distance;
516  if (min_sqr_distance > max_min_sqr_distance)
517  max_min_sqr_distance = min_sqr_distance;
518 
519  //std::cerr << min_sqr_distance << ", " << min_min_sqr_distance << ", " << max_min_sqr_distance << std::endl;
520  }
521 
522  if (best_iter != list1.end ())
523  {
524  //std::cerr << "feature_index: " << list2.size () << std::endl;
525  //std::cerr << "min_min_sqr_distance: " << min_min_sqr_distance << std::endl;
526  //std::cerr << "max_min_sqr_distance: " << max_min_sqr_distance << std::endl;
527 
528  if (min_min_sqr_distance < 50)
529  {
530  feature_selection_finished = true;
531  break;
532  }
533 
534  list2.push_back (*best_iter);
535  }
536  }
537  }
538  else
539  {
540  if (list1.size () <= nr_features)
541  {
542  for (auto iter1 = list1.begin (); iter1 != list1.end (); ++iter1)
543  {
544  QuantizedMultiModFeature feature;
545 
546  feature.x = iter1->x;
547  feature.y = iter1->y;
548  feature.modality_index = modality_index;
549  feature.quantized_value = filtered_quantized_color_gradients_ (iter1->x, iter1->y);
550 
551  features.push_back (feature);
552  }
553  return;
554  }
555 
556  list2.push_back (*(list1.begin ()));
557  while (list2.size () != nr_features)
558  {
559  float best_score = 0.0f;
560  auto best_iter = list1.end ();
561  for (auto iter1 = list1.begin (); iter1 != list1.end (); ++iter1)
562  {
563  // find smallest distance
564  float smallest_distance = std::numeric_limits<float>::max ();
565  for (auto iter2 = list2.begin (); iter2 != list2.end (); ++iter2)
566  {
567  const float dx = static_cast<float> (iter1->x) - static_cast<float> (iter2->x);
568  const float dy = static_cast<float> (iter1->y) - static_cast<float> (iter2->y);
569 
570  const float distance = dx*dx + dy*dy;
571 
572  if (distance < smallest_distance)
573  {
574  smallest_distance = distance;
575  }
576  }
577 
578  const float score = smallest_distance * iter1->gradient.magnitude;
579 
580  if (score > best_score)
581  {
582  best_score = score;
583  best_iter = iter1;
584  }
585  }
586 
587  if (best_iter != list1.end ())
588  {
589  list2.push_back (*best_iter);
590  }
591  else
592  {
593  break;
594  }
595  }
596  }
597  }
598  else if (feature_selection_method_ == MASK_BORDER_HIGH_GRADIENTS || feature_selection_method_ == MASK_BORDER_EQUALLY)
599  {
600  MaskMap eroded_mask;
601  erode (mask, eroded_mask);
602 
603  auto diff_mask = MaskMap::getDifferenceMask (mask, eroded_mask);
604 
605  for (std::size_t row_index = 0; row_index < height; ++row_index)
606  {
607  for (std::size_t col_index = 0; col_index < width; ++col_index)
608  {
609  if (diff_mask (col_index, row_index) != 0)
610  {
611  const GradientXY & gradient = color_gradients_ (col_index, row_index);
612  if ((feature_selection_method_ == MASK_BORDER_EQUALLY || gradient.magnitude > gradient_magnitude_threshold_feature_extraction_)
613  && filtered_quantized_color_gradients_ (col_index, row_index) != 0)
614  {
615  Candidate candidate;
616  candidate.gradient = gradient;
617  candidate.x = static_cast<int> (col_index);
618  candidate.y = static_cast<int> (row_index);
619 
620  list1.push_back (candidate);
621  }
622  }
623  }
624  }
625 
626  list1.sort();
627 
628  if (list1.size () <= nr_features)
629  {
630  for (auto iter1 = list1.begin (); iter1 != list1.end (); ++iter1)
631  {
632  QuantizedMultiModFeature feature;
633 
634  feature.x = iter1->x;
635  feature.y = iter1->y;
636  feature.modality_index = modality_index;
637  feature.quantized_value = filtered_quantized_color_gradients_ (iter1->x, iter1->y);
638 
639  features.push_back (feature);
640  }
641  return;
642  }
643 
644  std::size_t distance = list1.size () / nr_features + 1; // ???
645  while (list2.size () != nr_features)
646  {
647  const std::size_t sqr_distance = distance*distance;
648  for (auto iter1 = list1.begin (); iter1 != list1.end (); ++iter1)
649  {
650  bool candidate_accepted = true;
651 
652  for (auto iter2 = list2.begin (); iter2 != list2.end (); ++iter2)
653  {
654  const int dx = iter1->x - iter2->x;
655  const int dy = iter1->y - iter2->y;
656  const unsigned int tmp_distance = dx*dx + dy*dy;
657 
658  //if (tmp_distance < distance)
659  if (tmp_distance < sqr_distance)
660  {
661  candidate_accepted = false;
662  break;
663  }
664  }
665 
666  if (candidate_accepted)
667  list2.push_back (*iter1);
668 
669  if (list2.size () == nr_features)
670  break;
671  }
672  --distance;
673  }
674  }
675 
676  for (auto iter2 = list2.begin (); iter2 != list2.end (); ++iter2)
677  {
678  QuantizedMultiModFeature feature;
679 
680  feature.x = iter2->x;
681  feature.y = iter2->y;
682  feature.modality_index = modality_index;
683  feature.quantized_value = filtered_quantized_color_gradients_ (iter2->x, iter2->y);
684 
685  features.push_back (feature);
686  }
687 }
688 
689 //////////////////////////////////////////////////////////////////////////////////////////////
690 template <typename PointInT> void
692 extractAllFeatures (const MaskMap & mask, const std::size_t, const std::size_t modality_index,
693  std::vector<QuantizedMultiModFeature> & features) const
694 {
695  const std::size_t width = mask.getWidth ();
696  const std::size_t height = mask.getHeight ();
697 
698  std::list<Candidate> list1;
699  std::list<Candidate> list2;
700 
701 
702  for (std::size_t row_index = 0; row_index < height; ++row_index)
703  {
704  for (std::size_t col_index = 0; col_index < width; ++col_index)
705  {
706  if (mask (col_index, row_index) != 0)
707  {
708  const GradientXY & gradient = color_gradients_ (col_index, row_index);
709  if (gradient.magnitude > gradient_magnitude_threshold_feature_extraction_
710  && filtered_quantized_color_gradients_ (col_index, row_index) != 0)
711  {
712  Candidate candidate;
713  candidate.gradient = gradient;
714  candidate.x = static_cast<int> (col_index);
715  candidate.y = static_cast<int> (row_index);
716 
717  list1.push_back (candidate);
718  }
719  }
720  }
721  }
722 
723  list1.sort();
724 
725  for (auto iter1 = list1.begin (); iter1 != list1.end (); ++iter1)
726  {
727  QuantizedMultiModFeature feature;
728 
729  feature.x = iter1->x;
730  feature.y = iter1->y;
731  feature.modality_index = modality_index;
732  feature.quantized_value = filtered_quantized_color_gradients_ (iter1->x, iter1->y);
733 
734  features.push_back (feature);
735  }
736 }
737 
738 //////////////////////////////////////////////////////////////////////////////////////////////
739 template <typename PointInT>
740 void
743 {
744  const int width = cloud->width;
745  const int height = cloud->height;
746 
747  color_gradients_.resize (width*height);
748  color_gradients_.width = width;
749  color_gradients_.height = height;
750 
751  const float pi = std::tan (1.0f) * 2;
752  for (int row_index = 0; row_index < height-2; ++row_index)
753  {
754  for (int col_index = 0; col_index < width-2; ++col_index)
755  {
756  const int index0 = row_index*width+col_index;
757  const int index_c = row_index*width+col_index+2;
758  const int index_r = (row_index+2)*width+col_index;
759 
760  //const int index_d = (row_index+1)*width+col_index+1;
761 
762  const unsigned char r0 = (*cloud)[index0].r;
763  const unsigned char g0 = (*cloud)[index0].g;
764  const unsigned char b0 = (*cloud)[index0].b;
765 
766  const unsigned char r_c = (*cloud)[index_c].r;
767  const unsigned char g_c = (*cloud)[index_c].g;
768  const unsigned char b_c = (*cloud)[index_c].b;
769 
770  const unsigned char r_r = (*cloud)[index_r].r;
771  const unsigned char g_r = (*cloud)[index_r].g;
772  const unsigned char b_r = (*cloud)[index_r].b;
773 
774  const float r_dx = static_cast<float> (r_c) - static_cast<float> (r0);
775  const float g_dx = static_cast<float> (g_c) - static_cast<float> (g0);
776  const float b_dx = static_cast<float> (b_c) - static_cast<float> (b0);
777 
778  const float r_dy = static_cast<float> (r_r) - static_cast<float> (r0);
779  const float g_dy = static_cast<float> (g_r) - static_cast<float> (g0);
780  const float b_dy = static_cast<float> (b_r) - static_cast<float> (b0);
781 
782  const float sqr_mag_r = r_dx*r_dx + r_dy*r_dy;
783  const float sqr_mag_g = g_dx*g_dx + g_dy*g_dy;
784  const float sqr_mag_b = b_dx*b_dx + b_dy*b_dy;
785 
786  if (sqr_mag_r > sqr_mag_g && sqr_mag_r > sqr_mag_b)
787  {
788  GradientXY gradient;
789  gradient.magnitude = std::sqrt (sqr_mag_r);
790  gradient.angle = std::atan2 (r_dy, r_dx) * 180.0f / pi;
791  gradient.x = static_cast<float> (col_index);
792  gradient.y = static_cast<float> (row_index);
793 
794  color_gradients_ (col_index+1, row_index+1) = gradient;
795  }
796  else if (sqr_mag_g > sqr_mag_b)
797  {
798  GradientXY gradient;
799  gradient.magnitude = std::sqrt (sqr_mag_g);
800  gradient.angle = std::atan2 (g_dy, g_dx) * 180.0f / pi;
801  gradient.x = static_cast<float> (col_index);
802  gradient.y = static_cast<float> (row_index);
803 
804  color_gradients_ (col_index+1, row_index+1) = gradient;
805  }
806  else
807  {
808  GradientXY gradient;
809  gradient.magnitude = std::sqrt (sqr_mag_b);
810  gradient.angle = std::atan2 (b_dy, b_dx) * 180.0f / pi;
811  gradient.x = static_cast<float> (col_index);
812  gradient.y = static_cast<float> (row_index);
813 
814  color_gradients_ (col_index+1, row_index+1) = gradient;
815  }
816 
817  assert (color_gradients_ (col_index+1, row_index+1).angle >= -180 &&
818  color_gradients_ (col_index+1, row_index+1).angle <= 180);
819  }
820  }
821 
822  return;
823 }
824 
825 //////////////////////////////////////////////////////////////////////////////////////////////
826 template <typename PointInT>
827 void
830 {
831  const int width = cloud->width;
832  const int height = cloud->height;
833 
834  color_gradients_.resize (width*height);
835  color_gradients_.width = width;
836  color_gradients_.height = height;
837 
838  const float pi = tanf (1.0f) * 2.0f;
839  for (int row_index = 1; row_index < height-1; ++row_index)
840  {
841  for (int col_index = 1; col_index < width-1; ++col_index)
842  {
843  const int r7 = static_cast<int> ((*cloud)[(row_index-1)*width + (col_index-1)].r);
844  const int g7 = static_cast<int> ((*cloud)[(row_index-1)*width + (col_index-1)].g);
845  const int b7 = static_cast<int> ((*cloud)[(row_index-1)*width + (col_index-1)].b);
846  const int r8 = static_cast<int> ((*cloud)[(row_index-1)*width + (col_index)].r);
847  const int g8 = static_cast<int> ((*cloud)[(row_index-1)*width + (col_index)].g);
848  const int b8 = static_cast<int> ((*cloud)[(row_index-1)*width + (col_index)].b);
849  const int r9 = static_cast<int> ((*cloud)[(row_index-1)*width + (col_index+1)].r);
850  const int g9 = static_cast<int> ((*cloud)[(row_index-1)*width + (col_index+1)].g);
851  const int b9 = static_cast<int> ((*cloud)[(row_index-1)*width + (col_index+1)].b);
852  const int r4 = static_cast<int> ((*cloud)[(row_index)*width + (col_index-1)].r);
853  const int g4 = static_cast<int> ((*cloud)[(row_index)*width + (col_index-1)].g);
854  const int b4 = static_cast<int> ((*cloud)[(row_index)*width + (col_index-1)].b);
855  const int r6 = static_cast<int> ((*cloud)[(row_index)*width + (col_index+1)].r);
856  const int g6 = static_cast<int> ((*cloud)[(row_index)*width + (col_index+1)].g);
857  const int b6 = static_cast<int> ((*cloud)[(row_index)*width + (col_index+1)].b);
858  const int r1 = static_cast<int> ((*cloud)[(row_index+1)*width + (col_index-1)].r);
859  const int g1 = static_cast<int> ((*cloud)[(row_index+1)*width + (col_index-1)].g);
860  const int b1 = static_cast<int> ((*cloud)[(row_index+1)*width + (col_index-1)].b);
861  const int r2 = static_cast<int> ((*cloud)[(row_index+1)*width + (col_index)].r);
862  const int g2 = static_cast<int> ((*cloud)[(row_index+1)*width + (col_index)].g);
863  const int b2 = static_cast<int> ((*cloud)[(row_index+1)*width + (col_index)].b);
864  const int r3 = static_cast<int> ((*cloud)[(row_index+1)*width + (col_index+1)].r);
865  const int g3 = static_cast<int> ((*cloud)[(row_index+1)*width + (col_index+1)].g);
866  const int b3 = static_cast<int> ((*cloud)[(row_index+1)*width + (col_index+1)].b);
867 
868  //const int r_tmp1 = - r7 + r3;
869  //const int r_tmp2 = - r1 + r9;
870  //const int g_tmp1 = - g7 + g3;
871  //const int g_tmp2 = - g1 + g9;
872  //const int b_tmp1 = - b7 + b3;
873  //const int b_tmp2 = - b1 + b9;
874  ////const int gx = - r7 - (r4<<2) - r1 + r3 + (r6<<2) + r9;
875  ////const int gy = - r7 - (r8<<2) - r9 + r1 + (r2<<2) + r3;
876  //const int r_dx = r_tmp1 + r_tmp2 - (r4<<2) + (r6<<2);
877  //const int r_dy = r_tmp1 - r_tmp2 - (r8<<2) + (r2<<2);
878  //const int g_dx = g_tmp1 + g_tmp2 - (g4<<2) + (g6<<2);
879  //const int g_dy = g_tmp1 - g_tmp2 - (g8<<2) + (g2<<2);
880  //const int b_dx = b_tmp1 + b_tmp2 - (b4<<2) + (b6<<2);
881  //const int b_dy = b_tmp1 - b_tmp2 - (b8<<2) + (b2<<2);
882 
883  //const int r_tmp1 = - r7 + r3;
884  //const int r_tmp2 = - r1 + r9;
885  //const int g_tmp1 = - g7 + g3;
886  //const int g_tmp2 = - g1 + g9;
887  //const int b_tmp1 = - b7 + b3;
888  //const int b_tmp2 = - b1 + b9;
889  //const int gx = - r7 - (r4<<2) - r1 + r3 + (r6<<2) + r9;
890  //const int gy = - r7 - (r8<<2) - r9 + r1 + (r2<<2) + r3;
891  const int r_dx = r9 + 2*r6 + r3 - (r7 + 2*r4 + r1);
892  const int r_dy = r1 + 2*r2 + r3 - (r7 + 2*r8 + r9);
893  const int g_dx = g9 + 2*g6 + g3 - (g7 + 2*g4 + g1);
894  const int g_dy = g1 + 2*g2 + g3 - (g7 + 2*g8 + g9);
895  const int b_dx = b9 + 2*b6 + b3 - (b7 + 2*b4 + b1);
896  const int b_dy = b1 + 2*b2 + b3 - (b7 + 2*b8 + b9);
897 
898  const int sqr_mag_r = r_dx*r_dx + r_dy*r_dy;
899  const int sqr_mag_g = g_dx*g_dx + g_dy*g_dy;
900  const int sqr_mag_b = b_dx*b_dx + b_dy*b_dy;
901 
902  if (sqr_mag_r > sqr_mag_g && sqr_mag_r > sqr_mag_b)
903  {
904  GradientXY gradient;
905  gradient.magnitude = std::sqrt (static_cast<float> (sqr_mag_r));
906  gradient.angle = std::atan2 (static_cast<float> (r_dy), static_cast<float> (r_dx)) * 180.0f / pi;
907  if (gradient.angle < -180.0f) gradient.angle += 360.0f;
908  if (gradient.angle >= 180.0f) gradient.angle -= 360.0f;
909  gradient.x = static_cast<float> (col_index);
910  gradient.y = static_cast<float> (row_index);
911 
912  color_gradients_ (col_index, row_index) = gradient;
913  }
914  else if (sqr_mag_g > sqr_mag_b)
915  {
916  GradientXY gradient;
917  gradient.magnitude = std::sqrt (static_cast<float> (sqr_mag_g));
918  gradient.angle = std::atan2 (static_cast<float> (g_dy), static_cast<float> (g_dx)) * 180.0f / pi;
919  if (gradient.angle < -180.0f) gradient.angle += 360.0f;
920  if (gradient.angle >= 180.0f) gradient.angle -= 360.0f;
921  gradient.x = static_cast<float> (col_index);
922  gradient.y = static_cast<float> (row_index);
923 
924  color_gradients_ (col_index, row_index) = gradient;
925  }
926  else
927  {
928  GradientXY gradient;
929  gradient.magnitude = std::sqrt (static_cast<float> (sqr_mag_b));
930  gradient.angle = std::atan2 (static_cast<float> (b_dy), static_cast<float> (b_dx)) * 180.0f / pi;
931  if (gradient.angle < -180.0f) gradient.angle += 360.0f;
932  if (gradient.angle >= 180.0f) gradient.angle -= 360.0f;
933  gradient.x = static_cast<float> (col_index);
934  gradient.y = static_cast<float> (row_index);
935 
936  color_gradients_ (col_index, row_index) = gradient;
937  }
938 
939  assert (color_gradients_ (col_index, row_index).angle >= -180 &&
940  color_gradients_ (col_index, row_index).angle <= 180);
941  }
942  }
943 
944  return;
945 }
946 
947 //////////////////////////////////////////////////////////////////////////////////////////////
948 template <typename PointInT>
949 void
952 {
953  //std::cerr << "quantize this, bastard!!!" << std::endl;
954 
955  //unsigned char quantization_map[16] = {0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7};
956  //unsigned char quantization_map[16] = {1,2,3,4,5,6,7,8,1,2,3,4,5,6,7,8};
957 
958  //for (float angle = 0.0f; angle < 360.0f; angle += 1.0f)
959  //{
960  // const int quantized_value = quantization_map[static_cast<int> (angle * angleScale)];
961  // std::cerr << angle << ": " << quantized_value << std::endl;
962  //}
963 
964 
965  const std::size_t width = input_->width;
966  const std::size_t height = input_->height;
967 
968  quantized_color_gradients_.resize (width, height);
969 
970  constexpr float angleScale = 16.0f / 360.0f;
971 
972  //float min_angle = std::numeric_limits<float>::max ();
973  //float max_angle = -std::numeric_limits<float>::max ();
974  for (std::size_t row_index = 0; row_index < height; ++row_index)
975  {
976  for (std::size_t col_index = 0; col_index < width; ++col_index)
977  {
978  if (color_gradients_ (col_index, row_index).magnitude < gradient_magnitude_threshold_)
979  {
980  quantized_color_gradients_ (col_index, row_index) = 0;
981  continue;
982  }
983 
984  const float angle = 11.25f + color_gradients_ (col_index, row_index).angle + 180.0f;
985  const int quantized_value = (static_cast<int> (angle * angleScale)) & 7;
986  quantized_color_gradients_ (col_index, row_index) = static_cast<unsigned char> (quantized_value + 1);
987 
988  //const float angle = color_gradients_ (col_index, row_index).angle + 180.0f;
989 
990  //min_angle = std::min (min_angle, angle);
991  //max_angle = std::max (max_angle, angle);
992 
993  //if (angle < 0.0f || angle >= 360.0f)
994  //{
995  // std::cerr << "angle shitty: " << angle << std::endl;
996  //}
997 
998  //const int quantized_value = quantization_map[static_cast<int> (angle * angleScale)];
999  //quantized_color_gradients_ (col_index, row_index) = static_cast<unsigned char> (quantized_value);
1000 
1001  //assert (0 <= quantized_value && quantized_value < 16);
1002  //quantized_color_gradients_ (col_index, row_index) = quantization_map[quantized_value];
1003  //quantized_color_gradients_ (col_index, row_index) = static_cast<unsigned char> ((quantized_value & 7) + 1); // = (quantized_value % 8) + 1
1004  }
1005  }
1006 
1007  //std::cerr << ">>>>> " << min_angle << ", " << max_angle << std::endl;
1008 }
1009 
1010 //////////////////////////////////////////////////////////////////////////////////////////////
1011 template <typename PointInT>
1012 void
1015 {
1016  const std::size_t width = input_->width;
1017  const std::size_t height = input_->height;
1018 
1019  filtered_quantized_color_gradients_.resize (width, height);
1020 
1021  // filter data
1022  for (std::size_t row_index = 1; row_index < height-1; ++row_index)
1023  {
1024  for (std::size_t col_index = 1; col_index < width-1; ++col_index)
1025  {
1026  unsigned char histogram[9] = {0,0,0,0,0,0,0,0,0};
1027 
1028  {
1029  const unsigned char * data_ptr = quantized_color_gradients_.getData () + (row_index-1)*width+col_index-1;
1030  assert (data_ptr[0] < 9 && data_ptr[1] < 9 && data_ptr[2] < 9);
1031  ++histogram[data_ptr[0]];
1032  ++histogram[data_ptr[1]];
1033  ++histogram[data_ptr[2]];
1034  }
1035  {
1036  const unsigned char * data_ptr = quantized_color_gradients_.getData () + row_index*width+col_index-1;
1037  assert (data_ptr[0] < 9 && data_ptr[1] < 9 && data_ptr[2] < 9);
1038  ++histogram[data_ptr[0]];
1039  ++histogram[data_ptr[1]];
1040  ++histogram[data_ptr[2]];
1041  }
1042  {
1043  const unsigned char * data_ptr = quantized_color_gradients_.getData () + (row_index+1)*width+col_index-1;
1044  assert (data_ptr[0] < 9 && data_ptr[1] < 9 && data_ptr[2] < 9);
1045  ++histogram[data_ptr[0]];
1046  ++histogram[data_ptr[1]];
1047  ++histogram[data_ptr[2]];
1048  }
1049 
1050  unsigned char max_hist_value = 0;
1051  int max_hist_index = -1;
1052 
1053  // for (int i = 0; i < 8; ++i)
1054  // {
1055  // if (max_hist_value < histogram[i+1])
1056  // {
1057  // max_hist_index = i;
1058  // max_hist_value = histogram[i+1]
1059  // }
1060  // }
1061  // Unrolled for performance optimization:
1062  if (max_hist_value < histogram[1]) {max_hist_index = 0; max_hist_value = histogram[1];}
1063  if (max_hist_value < histogram[2]) {max_hist_index = 1; max_hist_value = histogram[2];}
1064  if (max_hist_value < histogram[3]) {max_hist_index = 2; max_hist_value = histogram[3];}
1065  if (max_hist_value < histogram[4]) {max_hist_index = 3; max_hist_value = histogram[4];}
1066  if (max_hist_value < histogram[5]) {max_hist_index = 4; max_hist_value = histogram[5];}
1067  if (max_hist_value < histogram[6]) {max_hist_index = 5; max_hist_value = histogram[6];}
1068  if (max_hist_value < histogram[7]) {max_hist_index = 6; max_hist_value = histogram[7];}
1069  if (max_hist_value < histogram[8]) {max_hist_index = 7; max_hist_value = histogram[8];}
1070 
1071  if (max_hist_index != -1 && max_hist_value >= 5)
1072  filtered_quantized_color_gradients_ (col_index, row_index) = static_cast<unsigned char> (0x1 << max_hist_index);
1073  else
1074  filtered_quantized_color_gradients_ (col_index, row_index) = 0;
1075 
1076  }
1077  }
1078 }
1079 
1080 //////////////////////////////////////////////////////////////////////////////////////////////
1081 template <typename PointInT>
1082 void
1084 erode (const pcl::MaskMap & mask_in,
1085  pcl::MaskMap & mask_out)
1086 {
1087  const std::size_t width = mask_in.getWidth ();
1088  const std::size_t height = mask_in.getHeight ();
1089 
1090  mask_out.resize (width, height);
1091 
1092  for (std::size_t row_index = 1; row_index < height-1; ++row_index)
1093  {
1094  for (std::size_t col_index = 1; col_index < width-1; ++col_index)
1095  {
1096  if (mask_in (col_index, row_index-1) == 0 ||
1097  mask_in (col_index-1, row_index) == 0 ||
1098  mask_in (col_index+1, row_index) == 0 ||
1099  mask_in (col_index, row_index+1) == 0)
1100  {
1101  mask_out (col_index, row_index) = 0;
1102  }
1103  else
1104  {
1105  mask_out (col_index, row_index) = 255;
1106  }
1107  }
1108  }
1109 }
QuantizedMap & getQuantizedMap() override
Returns a reference to the internally computed quantized map.
static void spreadQuantizedMap(const QuantizedMap &input_map, QuantizedMap &output_map, std::size_t spreading_size)
void extractAllFeatures(const MaskMap &mask, std::size_t nr_features, std::size_t modalityIndex, std::vector< QuantizedMultiModFeature > &features) const override
Extracts all possible features from the modality within the specified mask.
shared_ptr< PointCloud< PointT > > Ptr
Definition: point_cloud.h:413
Feature that defines a position and quantized value in a specific modality.
void computeMaxColorGradientsSobel(const typename pcl::PointCloud< pcl::RGB >::ConstPtr &cloud)
Computes the max-RGB gradients for the specified cloud using sobel.
void convolve(const Eigen::ArrayXf &h_kernel, const Eigen::ArrayXf &v_kernel, PointCloudOut &output)
Convolve point cloud with an horizontal kernel along rows then vertical kernel along columns : convol...
void quantizeColorGradients()
Quantizes the color gradients.
void setFeatureSelectionMethod(const FeatureSelectionMethod method)
Sets the feature selection method.
std::size_t modality_index
the index of the corresponding modality.
unsigned char quantized_value
the quantized value attached to the feature.
void setKernel(const Eigen::ArrayXf &kernel)
Set convolving kernel.
Definition: convolution.h:104
A point structure representing Euclidean xyz coordinates, and the intensity value.
Definition: point_types.h:52
void setSpreadingSize(const std::size_t spreading_size)
Sets the spreading size for spreading the quantized data.
std::uint32_t width
The point cloud width (if organized as an image-structure).
Definition: point_cloud.h:398
A structure representing RGB color information.
~ColorGradientModality() override
Destructor.
Convolution is a mathematical operation on two functions f and g, producing a third function that is ...
Definition: convolution.h:72
QuantizedMap & getSpreadedQuantizedMap() override
Returns a reference to the internally computed spread quantized map.
void setInputCloud(const PointCloudInConstPtr &cloud)
Provide a pointer to the input dataset.
Definition: convolution.h:99
PCL base class.
Definition: pcl_base.h:69
void setGradientMagnitudeThresholdForFeatureExtraction(const float threshold)
Sets the threshold for the gradient magnitude which is used for feature extraction.
void computeMaxColorGradients(const typename pcl::PointCloud< pcl::RGB >::ConstPtr &cloud)
Computes the max-RGB gradients for the specified cloud.
void setInputCloud(const typename PointCloudIn::ConstPtr &cloud) override
Provide a pointer to the input dataset (overwrites the PCLBase::setInputCloud method) ...
Interface for a quantizable modality.
void resize(std::size_t count)
Resizes the container to contain count elements.
Definition: point_cloud.h:462
std::uint32_t height
The point cloud height (if organized as an image-structure).
Definition: point_cloud.h:400
virtual void processInputDataFromFiltered()
Processes the input data assuming that everything up to filtering is already done/available (so only ...
void resize(std::size_t width, std::size_t height)
Modality based on max-RGB gradients.
void filterQuantizedColorGradients()
Filters the quantized gradients.
FeatureSelectionMethod
Different methods for feature selection/extraction.
void extractFeatures(const MaskMap &mask, std::size_t nr_features, std::size_t modalityIndex, std::vector< QuantizedMultiModFeature > &features) const override
Extracts features from this modality within the specified mask.
bool is_dense
True if no points are invalid (e.g., have NaN or Inf values in any of their floating point fields)...
Definition: point_cloud.h:403
static PCL_NODISCARD MaskMap getDifferenceMask(const MaskMap &mask0, const MaskMap &mask1)
shared_ptr< const PointCloud< PointRGBT > > ConstPtr
Definition: point_cloud.h:414
virtual void processInputData()
Processes the input data (smoothing, computing gradients, quantizing, filtering, spreading).
PointCloudConstPtr input_
The input point cloud dataset.
Definition: pcl_base.h:147
static void erode(const pcl::MaskMap &mask_in, pcl::MaskMap &mask_out)
Erodes a mask.
void setGradientMagnitudeThreshold(const float threshold)
Sets the threshold for the gradient magnitude which is used when quantizing the data.
bool operator<(const Candidate &rhs) const
Operator for comparing to candidates (by magnitude of the gradient).
void computeGaussianKernel(const std::size_t kernel_size, const float sigma, std::vector< float > &kernel_values)
Computes the Gaussian kernel used for smoothing.
void setVariableFeatureNr(const bool enabled)
Sets whether variable feature numbers for feature extraction is enabled.
pcl::PointCloud< pcl::GradientXY > & getMaxColorGradients()
Returns a point cloud containing the max-RGB gradients.
std::size_t getHeight() const
Definition: mask_map.h:63
Candidate for a feature (used in feature extraction methods).
std::size_t getWidth() const
Definition: mask_map.h:57