Point Cloud Library (PCL) 1.12.0
Loading...
Searching...
No Matches
spin_image.hpp
1/*
2 * Software License Agreement (BSD License)
3 *
4 * Point Cloud Library (PCL) - www.pointclouds.org
5 * Copyright (c) 2010-2012, Willow Garage, Inc.
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
41#ifndef PCL_FEATURES_IMPL_SPIN_IMAGE_H_
42#define PCL_FEATURES_IMPL_SPIN_IMAGE_H_
43
44#include <limits>
45#include <pcl/point_types.h>
46#include <pcl/exceptions.h>
47#include <pcl/features/spin_image.h>
48#include <cmath>
49
50//////////////////////////////////////////////////////////////////////////////////////////////
51template <typename PointInT, typename PointNT, typename PointOutT>
53 unsigned int image_width, double support_angle_cos, unsigned int min_pts_neighb) :
54 input_normals_ (), rotation_axes_cloud_ (),
55 is_angular_ (false), rotation_axis_ (), use_custom_axis_(false), use_custom_axes_cloud_ (false),
56 is_radial_ (false), image_width_ (image_width), support_angle_cos_ (support_angle_cos),
57 min_pts_neighb_ (min_pts_neighb)
58{
59 assert (support_angle_cos_ <= 1.0 && support_angle_cos_ >= 0.0); // may be permit negative cosine?
60
61 feature_name_ = "SpinImageEstimation";
62}
63
64
65//////////////////////////////////////////////////////////////////////////////////////////////
66template <typename PointInT, typename PointNT, typename PointOutT> Eigen::ArrayXXd
68{
69 assert (image_width_ > 0);
70 assert (support_angle_cos_ <= 1.0 && support_angle_cos_ >= 0.0); // may be permit negative cosine?
71
72 const Eigen::Vector3f origin_point ((*input_)[index].getVector3fMap ());
73
74 Eigen::Vector3f origin_normal;
75 origin_normal =
76 input_normals_ ?
77 (*input_normals_)[index].getNormalVector3fMap () :
78 Eigen::Vector3f (); // just a placeholder; should never be used!
79
80 const Eigen::Vector3f rotation_axis = use_custom_axis_ ?
81 rotation_axis_.getNormalVector3fMap () :
82 use_custom_axes_cloud_ ?
83 (*rotation_axes_cloud_)[index].getNormalVector3fMap () :
84 origin_normal;
85
86 Eigen::ArrayXXd m_matrix (Eigen::ArrayXXd::Zero (image_width_+1, 2*image_width_+1));
87 Eigen::ArrayXXd m_averAngles (Eigen::ArrayXXd::Zero (image_width_+1, 2*image_width_+1));
88
89 // OK, we are interested in the points of the cylinder of height 2*r and
90 // base radius r, where r = m_dBinSize * in_iImageWidth
91 // it can be embedded to the sphere of radius sqrt(2) * m_dBinSize * in_iImageWidth
92 // suppose that points are uniformly distributed, so we lose ~40%
93 // according to the volumes ratio
94 double bin_size = 0.0;
95 if (is_radial_)
96 bin_size = search_radius_ / image_width_;
97 else
98 bin_size = search_radius_ / image_width_ / sqrt(2.0);
99
100 pcl::Indices nn_indices;
101 std::vector<float> nn_sqr_dists;
102 const int neighb_cnt = this->searchForNeighbors (index, search_radius_, nn_indices, nn_sqr_dists);
103 if (neighb_cnt < static_cast<int> (min_pts_neighb_))
104 {
105 throw PCLException (
106 "Too few points for spin image, use setMinPointCountInNeighbourhood() to decrease the threshold or use larger feature radius",
107 "spin_image.hpp", "computeSiForPoint");
108 }
109
110 // for all neighbor points
111 for (int i_neigh = 0; i_neigh < neighb_cnt ; i_neigh++)
112 {
113 // first, skip the points with distant normals
114 double cos_between_normals = -2.0; // should be initialized if used
115 if (support_angle_cos_ > 0.0 || is_angular_) // not bogus
116 {
117 cos_between_normals = origin_normal.dot ((*input_normals_)[nn_indices[i_neigh]].getNormalVector3fMap ());
118 if (std::abs (cos_between_normals) > (1.0 + 10*std::numeric_limits<float>::epsilon ())) // should be okay for numeric stability
119 {
120 PCL_ERROR ("[pcl::%s::computeSiForPoint] Normal for the point %d and/or the point %d are not normalized, dot ptoduct is %f.\n",
121 getClassName ().c_str (), nn_indices[i_neigh], index, cos_between_normals);
122 throw PCLException ("Some normals are not normalized",
123 "spin_image.hpp", "computeSiForPoint");
124 }
125 cos_between_normals = std::max (-1.0, std::min (1.0, cos_between_normals));
126
127 if (std::abs (cos_between_normals) < support_angle_cos_ ) // allow counter-directed normals
128 {
129 continue;
130 }
131
132 if (cos_between_normals < 0.0)
133 {
134 cos_between_normals = -cos_between_normals; // the normal is not used explicitly from now
135 }
136 }
137
138 // now compute the coordinate in cylindric coordinate system associated with the origin point
139 const Eigen::Vector3f direction (
140 (*surface_)[nn_indices[i_neigh]].getVector3fMap () - origin_point);
141 const double direction_norm = direction.norm ();
142 if (std::abs(direction_norm) < 10*std::numeric_limits<double>::epsilon ())
143 continue; // ignore the point itself; it does not contribute really
144 assert (direction_norm > 0.0);
145
146 // the angle between the normal vector and the direction to the point
147 double cos_dir_axis = direction.dot(rotation_axis) / direction_norm;
148 if (std::abs(cos_dir_axis) > (1.0 + 10*std::numeric_limits<float>::epsilon())) // should be okay for numeric stability
149 {
150 PCL_ERROR ("[pcl::%s::computeSiForPoint] Rotation axis for the point %d are not normalized, dot ptoduct is %f.\n",
151 getClassName ().c_str (), index, cos_dir_axis);
152 throw PCLException ("Some rotation axis is not normalized",
153 "spin_image.hpp", "computeSiForPoint");
154 }
155 cos_dir_axis = std::max (-1.0, std::min (1.0, cos_dir_axis));
156
157 // compute coordinates w.r.t. the reference frame
158 double beta = std::numeric_limits<double>::signaling_NaN ();
159 double alpha = std::numeric_limits<double>::signaling_NaN ();
160 if (is_radial_) // radial spin image structure
161 {
162 beta = asin (cos_dir_axis); // yes, arc sine! to get the angle against tangent, not normal!
163 alpha = direction_norm;
164 }
165 else // rectangular spin-image structure
166 {
167 beta = direction_norm * cos_dir_axis;
168 alpha = direction_norm * sqrt (1.0 - cos_dir_axis*cos_dir_axis);
169
170 if (std::abs (beta) >= bin_size * image_width_ || alpha >= bin_size * image_width_)
171 {
172 continue; // outside the cylinder
173 }
174 }
175
176 assert (alpha >= 0.0);
177 assert (alpha <= bin_size * image_width_ + 20 * std::numeric_limits<float>::epsilon () );
178
179
180 // bilinear interpolation
181 double beta_bin_size = is_radial_ ? (M_PI / 2 / image_width_) : bin_size;
182 int beta_bin = int(std::floor (beta / beta_bin_size)) + int(image_width_);
183 assert (0 <= beta_bin && beta_bin < m_matrix.cols ());
184 int alpha_bin = int(std::floor (alpha / bin_size));
185 assert (0 <= alpha_bin && alpha_bin < m_matrix.rows ());
186
187 if (alpha_bin == static_cast<int> (image_width_)) // border points
188 {
189 alpha_bin--;
190 // HACK: to prevent a > 1
191 alpha = bin_size * (alpha_bin + 1) - std::numeric_limits<double>::epsilon ();
192 }
193 if (beta_bin == int(2*image_width_) ) // border points
194 {
195 beta_bin--;
196 // HACK: to prevent b > 1
197 beta = beta_bin_size * (beta_bin - int(image_width_) + 1) - std::numeric_limits<double>::epsilon ();
198 }
199
200 double a = alpha/bin_size - double(alpha_bin);
201 double b = beta/beta_bin_size - double(beta_bin-int(image_width_));
202
203 assert (0 <= a && a <= 1);
204 assert (0 <= b && b <= 1);
205
206 m_matrix (alpha_bin, beta_bin) += (1-a) * (1-b);
207 m_matrix (alpha_bin+1, beta_bin) += a * (1-b);
208 m_matrix (alpha_bin, beta_bin+1) += (1-a) * b;
209 m_matrix (alpha_bin+1, beta_bin+1) += a * b;
210
211 if (is_angular_)
212 {
213 m_averAngles (alpha_bin, beta_bin) += (1-a) * (1-b) * std::acos (cos_between_normals);
214 m_averAngles (alpha_bin+1, beta_bin) += a * (1-b) * std::acos (cos_between_normals);
215 m_averAngles (alpha_bin, beta_bin+1) += (1-a) * b * std::acos (cos_between_normals);
216 m_averAngles (alpha_bin+1, beta_bin+1) += a * b * std::acos (cos_between_normals);
217 }
218 }
219
220 if (is_angular_)
221 {
222 // transform sum to average
223 m_matrix = m_averAngles / (m_matrix + std::numeric_limits<double>::epsilon ()); // +eps to avoid division by zero
224 }
225 else if (neighb_cnt > 1) // to avoid division by zero, also no need to divide by 1
226 {
227 // normalization
228 m_matrix /= m_matrix.sum();
229 }
230
231 return m_matrix;
232}
233
234
235//////////////////////////////////////////////////////////////////////////////////////////////
236template <typename PointInT, typename PointNT, typename PointOutT> bool
238{
240 {
241 PCL_ERROR ("[pcl::%s::initCompute] Init failed.\n", getClassName ().c_str ());
242 return (false);
243 }
244
245 // Check if input normals are set
246 if (!input_normals_)
247 {
248 PCL_ERROR ("[pcl::%s::initCompute] No input dataset containing normals was given!\n", getClassName ().c_str ());
250 return (false);
251 }
252
253 // Check if the size of normals is the same as the size of the surface
254 if (input_normals_->size () != input_->size ())
255 {
256 PCL_ERROR ("[pcl::%s::initCompute] ", getClassName ().c_str ());
257 PCL_ERROR ("The number of points in the input dataset differs from ");
258 PCL_ERROR ("the number of points in the dataset containing the normals!\n");
260 return (false);
261 }
262
263 // We need a positive definite search radius to continue
264 if (search_radius_ == 0)
265 {
266 PCL_ERROR ("[pcl::%s::initCompute] Need a search radius different than 0!\n", getClassName ().c_str ());
268 return (false);
269 }
270 if (k_ != 0)
271 {
272 PCL_ERROR ("[pcl::%s::initCompute] K-nearest neighbor search for spin images not implemented. Used a search radius instead!\n", getClassName ().c_str ());
274 return (false);
275 }
276 // If the surface won't be set, make fake surface and fake surface normals
277 // if we wouldn't do it here, the following method would alarm that no surface normals is given
278 if (!surface_)
279 {
280 surface_ = input_;
281 fake_surface_ = true;
282 }
283
284 //if (fake_surface_ && !input_normals_)
285 // input_normals_ = normals_; // normals_ is set, as checked earlier
286
287 assert(!(use_custom_axis_ && use_custom_axes_cloud_));
288
289 if (!use_custom_axis_ && !use_custom_axes_cloud_ // use input normals as rotation axes
290 && !input_normals_)
291 {
292 PCL_ERROR ("[pcl::%s::initCompute] No normals for input cloud were given!\n", getClassName ().c_str ());
293 // Cleanup
295 return (false);
296 }
297
298 if ((is_angular_ || support_angle_cos_ > 0.0) // support angle is not bogus NOTE this is for randomly-flipped normals
299 && !input_normals_)
300 {
301 PCL_ERROR ("[pcl::%s::initCompute] No normals for input cloud were given!\n", getClassName ().c_str ());
302 // Cleanup
304 return (false);
305 }
306
307 if (use_custom_axes_cloud_
308 && rotation_axes_cloud_->size () == input_->size ())
309 {
310 PCL_ERROR ("[pcl::%s::initCompute] Rotation axis cloud have different size from input!\n", getClassName ().c_str ());
311 // Cleanup
313 return (false);
314 }
315
316 return (true);
317}
318
319
320//////////////////////////////////////////////////////////////////////////////////////////////
321template <typename PointInT, typename PointNT, typename PointOutT> void
323{
324 for (std::size_t i_input = 0; i_input < indices_->size (); ++i_input)
325 {
326 Eigen::ArrayXXd res = computeSiForPoint (indices_->at (i_input));
327
328 // Copy into the resultant cloud
329 for (Eigen::Index iRow = 0; iRow < res.rows () ; iRow++)
330 {
331 for (Eigen::Index iCol = 0; iCol < res.cols () ; iCol++)
332 {
333 output[i_input].histogram[ iRow*res.cols () + iCol ] = static_cast<float> (res (iRow, iCol));
334 }
335 }
336 }
337}
338
339#define PCL_INSTANTIATE_SpinImageEstimation(T,NT,OutT) template class PCL_EXPORTS pcl::SpinImageEstimation<T,NT,OutT>;
340
341#endif // PCL_FEATURES_IMPL_SPIN_IMAGE_H_
342
Feature represents the base feature class.
Definition feature.h:107
std::string feature_name_
The feature name.
Definition feature.h:223
virtual bool deinitCompute()
This method should get called after ending the actual computation.
Definition feature.hpp:181
A base class for all pcl exceptions which inherits from std::runtime_error.
Definition exceptions.h:64
typename Feature< PointInT, PointOutT >::PointCloudOut PointCloudOut
Definition spin_image.h:101
Eigen::ArrayXXd computeSiForPoint(int index) const
Computes a spin-image for the point of the scan.
SpinImageEstimation(unsigned int image_width=8, double support_angle_cos=0.0, unsigned int min_pts_neighb=0)
Constructs empty spin image estimator.
bool initCompute() override
initializes computations specific to spin-image.
void computeFeature(PointCloudOut &output) override
Estimate the Spin Image descriptors at a set of points given by setInputWithNormals() using the surfa...
Defines all the PCL implemented PointT point type structures.
IndicesAllocator<> Indices
Type used for indices in PCL.
Definition types.h:133
#define M_PI
Definition pcl_macros.h:201