Point Cloud Library (PCL) 1.13.0
multiscale_feature_persistence.hpp
1/*
2 * Software License Agreement (BSD License)
3 *
4 * Point Cloud Library (PCL) - www.pointclouds.org
5 * Copyright (c) 2011, Alexandru-Eugen Ichim
6 * Copyright (c) 2012-, Open Perception, Inc.
7 *
8 * All rights reserved.
9 *
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
12 * are met:
13 *
14 * * Redistributions of source code must retain the above copyright
15 * notice, this list of conditions and the following disclaimer.
16 * * Redistributions in binary form must reproduce the above
17 * copyright notice, this list of conditions and the following
18 * disclaimer in the documentation and/or other materials provided
19 * with the distribution.
20 * * Neither the name of the copyright holder(s) nor the names of its
21 * contributors may be used to endorse or promote products derived
22 * from this software without specific prior written permission.
23 *
24 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
25 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
26 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
27 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
28 * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
29 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
30 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
31 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
32 * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
33 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
34 * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
35 * POSSIBILITY OF SUCH DAMAGE.
36 *
37 * $Id$
38 */
39
40#ifndef PCL_FEATURES_IMPL_MULTISCALE_FEATURE_PERSISTENCE_H_
41#define PCL_FEATURES_IMPL_MULTISCALE_FEATURE_PERSISTENCE_H_
42
43#include <pcl/features/multiscale_feature_persistence.h>
44
45//////////////////////////////////////////////////////////////////////////////////////////////
46template <typename PointSource, typename PointFeature>
48 alpha_ (0),
49 distance_metric_ (L1),
50 feature_estimator_ (),
51 features_at_scale_ (),
52 feature_representation_ ()
53{
54 feature_representation_.reset (new DefaultPointRepresentation<PointFeature>);
55 // No input is needed, hack around the initCompute () check from PCLBase
57}
58
59
60//////////////////////////////////////////////////////////////////////////////////////////////
61template <typename PointSource, typename PointFeature> bool
63{
65 {
66 PCL_ERROR ("[pcl::MultiscaleFeaturePersistence::initCompute] PCLBase::initCompute () failed - no input cloud was given.\n");
67 return false;
68 }
69 if (!feature_estimator_)
70 {
71 PCL_ERROR ("[pcl::MultiscaleFeaturePersistence::initCompute] No feature estimator was set\n");
72 return false;
73 }
74 if (scale_values_.empty ())
75 {
76 PCL_ERROR ("[pcl::MultiscaleFeaturePersistence::initCompute] No scale values were given\n");
77 return false;
78 }
79
80 mean_feature_.resize (feature_representation_->getNumberOfDimensions ());
81
82 return true;
83}
84
85
86//////////////////////////////////////////////////////////////////////////////////////////////
87template <typename PointSource, typename PointFeature> void
89{
90 features_at_scale_.clear ();
91 features_at_scale_.reserve (scale_values_.size ());
92 features_at_scale_vectorized_.clear ();
93 features_at_scale_vectorized_.reserve (scale_values_.size ());
94 for (std::size_t scale_i = 0; scale_i < scale_values_.size (); ++scale_i)
95 {
96 FeatureCloudPtr feature_cloud (new FeatureCloud ());
97 computeFeatureAtScale (scale_values_[scale_i], feature_cloud);
98 features_at_scale_.push_back(feature_cloud);
99
100 // Vectorize each feature and insert it into the vectorized feature storage
101 std::vector<std::vector<float> > feature_cloud_vectorized;
102 feature_cloud_vectorized.reserve (feature_cloud->size ());
103
104 for (const auto& feature: feature_cloud->points)
105 {
106 std::vector<float> feature_vectorized (feature_representation_->getNumberOfDimensions ());
107 feature_representation_->vectorize (feature, feature_vectorized);
108 feature_cloud_vectorized.emplace_back (std::move(feature_vectorized));
109 }
110 features_at_scale_vectorized_.emplace_back (std::move(feature_cloud_vectorized));
111 }
112}
113
114
115//////////////////////////////////////////////////////////////////////////////////////////////
116template <typename PointSource, typename PointFeature> void
118 FeatureCloudPtr &features)
119{
120 feature_estimator_->setRadiusSearch (scale);
121 feature_estimator_->compute (*features);
122}
123
124
125//////////////////////////////////////////////////////////////////////////////////////////////
126template <typename PointSource, typename PointFeature> float
128 const std::vector<float> &b)
129{
130 return (pcl::selectNorm<std::vector<float> > (a, b, a.size (), distance_metric_));
131}
132
133
134//////////////////////////////////////////////////////////////////////////////////////////////
135template <typename PointSource, typename PointFeature> void
137{
138 // Reset mean feature
139 std::fill_n(mean_feature_.begin (), mean_feature_.size (), 0.f);
140
141 std::size_t normalization_factor = 0;
142 for (const auto& scale: features_at_scale_vectorized_)
143 {
144 normalization_factor += scale.size (); // not using accumulate for cache efficiency
145 for (const auto &feature : scale)
146 std::transform(mean_feature_.cbegin (), mean_feature_.cend (),
147 feature.cbegin (), mean_feature_.begin (), std::plus<>{});
148 }
149
150 const float factor = std::max<float>(1, normalization_factor);
151 std::transform(mean_feature_.cbegin(),
152 mean_feature_.cend(),
153 mean_feature_.begin(),
154 [factor](const auto& mean) {
155 return mean / factor;
156 });
157}
158
159
160//////////////////////////////////////////////////////////////////////////////////////////////
161template <typename PointSource, typename PointFeature> void
163{
164 unique_features_indices_.clear ();
165 unique_features_table_.clear ();
166 unique_features_indices_.reserve (scale_values_.size ());
167 unique_features_table_.reserve (scale_values_.size ());
168
169 std::vector<float> diff_vector;
170 std::size_t size = 0;
171 for (const auto& feature : features_at_scale_vectorized_)
172 {
173 size = std::max(size, feature.size());
174 }
175 diff_vector.reserve(size);
176 for (std::size_t scale_i = 0; scale_i < features_at_scale_vectorized_.size (); ++scale_i)
177 {
178 // Calculate standard deviation within the scale
179 float standard_dev = 0.0;
180 diff_vector.clear();
181
182 for (const auto& feature: features_at_scale_vectorized_[scale_i])
183 {
184 float diff = distanceBetweenFeatures (feature, mean_feature_);
185 standard_dev += diff * diff;
186 diff_vector.emplace_back (diff);
187 }
188 standard_dev = std::sqrt (standard_dev / static_cast<float> (features_at_scale_vectorized_[scale_i].size ()));
189 PCL_DEBUG ("[pcl::MultiscaleFeaturePersistence::extractUniqueFeatures] Standard deviation for scale %f is %f\n", scale_values_[scale_i], standard_dev);
190
191 // Select only points outside (mean +/- alpha * standard_dev)
192 std::list<std::size_t> indices_per_scale;
193 std::vector<bool> indices_table_per_scale (features_at_scale_vectorized_[scale_i].size (), false);
194 for (std::size_t point_i = 0; point_i < features_at_scale_vectorized_[scale_i].size (); ++point_i)
195 {
196 if (diff_vector[point_i] > alpha_ * standard_dev)
197 {
198 indices_per_scale.emplace_back (point_i);
199 indices_table_per_scale[point_i] = true;
200 }
201 }
202 unique_features_indices_.emplace_back (std::move(indices_per_scale));
203 unique_features_table_.emplace_back (std::move(indices_table_per_scale));
204 }
205}
206
207
208//////////////////////////////////////////////////////////////////////////////////////////////
209template <typename PointSource, typename PointFeature> void
211 pcl::IndicesPtr &output_indices)
212{
213 if (!initCompute ())
214 return;
215
216 // Compute the features for all scales with the given feature estimator
217 PCL_DEBUG ("[pcl::MultiscaleFeaturePersistence::determinePersistentFeatures] Computing features ...\n");
218 computeFeaturesAtAllScales ();
219
220 // Compute mean feature
221 PCL_DEBUG ("[pcl::MultiscaleFeaturePersistence::determinePersistentFeatures] Calculating mean feature ...\n");
222 calculateMeanFeature ();
223
224 // Get the 'unique' features at each scale
225 PCL_DEBUG ("[pcl::MultiscaleFeaturePersistence::determinePersistentFeatures] Extracting unique features ...\n");
226 extractUniqueFeatures ();
227
228 PCL_DEBUG ("[pcl::MultiscaleFeaturePersistence::determinePersistentFeatures] Determining persistent features between scales ...\n");
229 // Determine persistent features between scales
230
231/*
232 // Method 1: a feature is considered persistent if it is 'unique' in at least 2 different scales
233 for (std::size_t scale_i = 0; scale_i < features_at_scale_vectorized_.size () - 1; ++scale_i)
234 for (std::list<std::size_t>::iterator feature_it = unique_features_indices_[scale_i].begin (); feature_it != unique_features_indices_[scale_i].end (); ++feature_it)
235 {
236 if (unique_features_table_[scale_i][*feature_it] == true)
237 {
238 output_features.push_back ((*features_at_scale_[scale_i])[*feature_it]);
239 output_indices->push_back (feature_estimator_->getIndices ()->at (*feature_it));
240 }
241 }
242*/
243 // Method 2: a feature is considered persistent if it is 'unique' in all the scales
244 for (const auto& feature: unique_features_indices_.front ())
245 {
246 bool present_in_all = true;
247 for (std::size_t scale_i = 0; scale_i < features_at_scale_.size (); ++scale_i)
248 present_in_all = present_in_all && unique_features_table_[scale_i][feature];
249
250 if (present_in_all)
251 {
252 output_features.emplace_back ((*features_at_scale_.front ())[feature]);
253 output_indices->emplace_back (feature_estimator_->getIndices ()->at (feature));
254 }
255 }
256
257 // Consider that output cloud is unorganized
258 output_features.header = feature_estimator_->getInputCloud ()->header;
259 output_features.is_dense = feature_estimator_->getInputCloud ()->is_dense;
260 output_features.width = output_features.size ();
261 output_features.height = 1;
262}
263
264
265#define PCL_INSTANTIATE_MultiscaleFeaturePersistence(InT, Feature) template class PCL_EXPORTS pcl::MultiscaleFeaturePersistence<InT, Feature>;
266
267#endif /* PCL_FEATURES_IMPL_MULTISCALE_FEATURE_PERSISTENCE_H_ */
DefaultPointRepresentation extends PointRepresentation to define default behavior for common point ty...
Generic class for extracting the persistent features from an input point cloud It can be given any Fe...
void determinePersistentFeatures(FeatureCloud &output_features, pcl::IndicesPtr &output_indices)
Central function that computes the persistent features.
void computeFeaturesAtAllScales()
Method that calls computeFeatureAtScale () for each scale parameter.
typename pcl::PointCloud< PointFeature >::Ptr FeatureCloudPtr
PCL base class.
Definition: pcl_base.h:70
PointCloudConstPtr input_
The input point cloud dataset.
Definition: pcl_base.h:147
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
std::uint32_t width
The point cloud width (if organized as an image-structure).
Definition: point_cloud.h:398
reference emplace_back(Args &&...args)
Emplace a new point in the cloud, at the end of the container.
Definition: point_cloud.h:686
pcl::PCLHeader header
The point cloud header.
Definition: point_cloud.h:392
std::uint32_t height
The point cloud height (if organized as an image-structure).
Definition: point_cloud.h:400
std::size_t size() const
Definition: point_cloud.h:443
float selectNorm(FloatVectorT a, FloatVectorT b, int dim, NormType norm_type)
Method that calculates any norm type available, based on the norm_type variable.
Definition: norms.hpp:50
@ L1
Definition: norms.h:54
shared_ptr< Indices > IndicesPtr
Definition: pcl_base.h:58