Point Cloud Library (PCL) 1.12.1
sac_model_normal_plane.hpp
1/*
2 * Software License Agreement (BSD License)
3 *
4 * Point Cloud Library (PCL) - www.pointclouds.org
5 * Copyright (c) 2009-2010, 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_SAMPLE_CONSENSUS_IMPL_SAC_MODEL_NORMAL_PLANE_H_
42#define PCL_SAMPLE_CONSENSUS_IMPL_SAC_MODEL_NORMAL_PLANE_H_
43
44#include <pcl/sample_consensus/sac_model_normal_plane.h>
45#include <pcl/sample_consensus/impl/sac_model_plane.hpp> // for dist4, dist8
46#include <pcl/common/common.h> // for getAngle3D
47
48//////////////////////////////////////////////////////////////////////////////////////////////////////////////////
49template <typename PointT, typename PointNT> void
51 const Eigen::VectorXf &model_coefficients, const double threshold, Indices &inliers)
52{
53 if (!normals_)
54 {
55 PCL_ERROR ("[pcl::SampleConsensusModelNormalPlane::selectWithinDistance] No input dataset containing normals was given!\n");
56 inliers.clear ();
57 return;
58 }
59
60 // Check if the model is valid given the user constraints
61 if (!isModelValid (model_coefficients))
62 {
63 inliers.clear ();
64 return;
65 }
66
67 // Obtain the plane normal
68 Eigen::Vector4f coeff = model_coefficients;
69 coeff[3] = 0.0f;
70
71 inliers.clear ();
72 error_sqr_dists_.clear ();
73 inliers.reserve (indices_->size ());
74 error_sqr_dists_.reserve (indices_->size ());
75
76 // Iterate through the 3d points and calculate the distances from them to the plane
77 for (std::size_t i = 0; i < indices_->size (); ++i)
78 {
79 const PointT &pt = (*input_)[(*indices_)[i]];
80 const PointNT &nt = (*normals_)[(*indices_)[i]];
81 // Calculate the distance from the point to the plane normal as the dot product
82 // D = (P-A).N/|N|
83 Eigen::Vector4f p (pt.x, pt.y, pt.z, 0.0f);
84 Eigen::Vector4f n (nt.normal_x, nt.normal_y, nt.normal_z, 0.0f);
85 double d_euclid = std::abs (coeff.dot (p) + model_coefficients[3]);
86
87 // Calculate the angular distance between the point normal and the plane normal
88 double d_normal = std::abs (getAngle3D (n, coeff));
89 d_normal = (std::min) (d_normal, M_PI - d_normal);
90
91 // Weight with the point curvature. On flat surfaces, curvature -> 0, which means the normal will have a higher influence
92 double weight = normal_distance_weight_ * (1.0 - nt.curvature);
93
94 double distance = std::abs (weight * d_normal + (1.0 - weight) * d_euclid);
95 if (distance < threshold)
96 {
97 // Returns the indices of the points whose distances are smaller than the threshold
98 inliers.push_back ((*indices_)[i]);
99 error_sqr_dists_.push_back (distance);
100 }
101 }
102}
103
104//////////////////////////////////////////////////////////////////////////////////////////////////////////////////
105template <typename PointT, typename PointNT> std::size_t
107 const Eigen::VectorXf &model_coefficients, const double threshold) const
108{
109 if (!normals_)
110 {
111 PCL_ERROR ("[pcl::SampleConsensusModelNormalPlane::countWithinDistance] No input dataset containing normals was given!\n");
112 return (0);
113 }
114
115 // Check if the model is valid given the user constraints
116 if (!isModelValid (model_coefficients))
117 return (0);
118
119#if defined (__AVX__) && defined (__AVX2__)
120 return countWithinDistanceAVX (model_coefficients, threshold);
121#elif defined (__SSE__) && defined (__SSE2__) && defined (__SSE4_1__)
122 return countWithinDistanceSSE (model_coefficients, threshold);
123#else
124 return countWithinDistanceStandard (model_coefficients, threshold);
125#endif
126}
127
128//////////////////////////////////////////////////////////////////////////
129template <typename PointT, typename PointNT> std::size_t
131 const Eigen::VectorXf &model_coefficients, const double threshold, std::size_t i) const
132{
133 std::size_t nr_p = 0;
134
135 // Obtain the plane normal
136 Eigen::Vector4f coeff = model_coefficients;
137 coeff[3] = 0.0f;
138
139 // Iterate through the 3d points and calculate the distances from them to the plane
140 for (; i < indices_->size (); ++i)
141 {
142 const PointT &pt = (*input_)[(*indices_)[i]];
143 const PointNT &nt = (*normals_)[(*indices_)[i]];
144 // Calculate the distance from the point to the plane normal as the dot product
145 // D = (P-A).N/|N|
146 const Eigen::Vector4f p (pt.x, pt.y, pt.z, 0.0f);
147 const Eigen::Vector4f n (nt.normal_x, nt.normal_y, nt.normal_z, 0.0f);
148 const double d_euclid = std::abs (coeff.dot (p) + model_coefficients[3]);
149
150 // Calculate the angular distance between the point normal and the plane normal
151 double d_normal = std::abs (getAngle3D (n, coeff));
152 d_normal = (std::min) (d_normal, M_PI - d_normal);
153
154 // Weight with the point curvature. On flat surfaces, curvature -> 0, which means the normal will have a higher influence
155 const double weight = normal_distance_weight_ * (1.0 - nt.curvature);
156
157 if (std::abs (weight * d_normal + (1.0 - weight) * d_euclid) < threshold)
158 {
159 nr_p++;
160 }
161 }
162 return (nr_p);
163}
164
165//////////////////////////////////////////////////////////////////////////
166#if defined (__SSE__) && defined (__SSE2__) && defined (__SSE4_1__)
167template <typename PointT, typename PointNT> std::size_t
169 const Eigen::VectorXf &model_coefficients, const double threshold, std::size_t i) const
170{
171 std::size_t nr_p = 0;
172 const __m128 a_vec = _mm_set1_ps (model_coefficients[0]);
173 const __m128 b_vec = _mm_set1_ps (model_coefficients[1]);
174 const __m128 c_vec = _mm_set1_ps (model_coefficients[2]);
175 const __m128 d_vec = _mm_set1_ps (model_coefficients[3]);
176 const __m128 threshold_vec = _mm_set1_ps (threshold);
177 const __m128 normal_distance_weight_vec = _mm_set1_ps (normal_distance_weight_);
178 const __m128 abs_help = _mm_set1_ps (-0.0F); // -0.0F (negative zero) means that all bits are 0, only the sign bit is 1
179 __m128i res = _mm_set1_epi32(0); // This corresponds to nr_p: 4 32bit integers that, summed together, hold the number of inliers
180 for (; (i + 4) <= indices_->size (); i += 4)
181 {
182 const __m128 d_euclid_vec = pcl::SampleConsensusModelPlane<PointT>::dist4 (i, a_vec, b_vec, c_vec, d_vec, abs_help);
183
184 const __m128 d_normal_vec = getAcuteAngle3DSSE (
185 _mm_set_ps ((*normals_)[(*indices_)[i ]].normal_x,
186 (*normals_)[(*indices_)[i+1]].normal_x,
187 (*normals_)[(*indices_)[i+2]].normal_x,
188 (*normals_)[(*indices_)[i+3]].normal_x),
189 _mm_set_ps ((*normals_)[(*indices_)[i ]].normal_y,
190 (*normals_)[(*indices_)[i+1]].normal_y,
191 (*normals_)[(*indices_)[i+2]].normal_y,
192 (*normals_)[(*indices_)[i+3]].normal_y),
193 _mm_set_ps ((*normals_)[(*indices_)[i ]].normal_z,
194 (*normals_)[(*indices_)[i+1]].normal_z,
195 (*normals_)[(*indices_)[i+2]].normal_z,
196 (*normals_)[(*indices_)[i+3]].normal_z),
197 a_vec, b_vec, c_vec);
198 const __m128 weight_vec = _mm_mul_ps (normal_distance_weight_vec, _mm_sub_ps (_mm_set1_ps (1.0f),
199 _mm_set_ps ((*normals_)[(*indices_)[i ]].curvature,
200 (*normals_)[(*indices_)[i+1]].curvature,
201 (*normals_)[(*indices_)[i+2]].curvature,
202 (*normals_)[(*indices_)[i+3]].curvature)));
203 const __m128 dist = _mm_andnot_ps (abs_help, _mm_add_ps (_mm_mul_ps (weight_vec, d_normal_vec), _mm_mul_ps (_mm_sub_ps (_mm_set1_ps (1.0f), weight_vec), d_euclid_vec)));
204 const __m128 mask = _mm_cmplt_ps (dist, threshold_vec); // The mask contains 1 bits if the corresponding points are inliers, else 0 bits
205 res = _mm_add_epi32 (res, _mm_and_si128 (_mm_set1_epi32 (1), _mm_castps_si128 (mask))); // The latter part creates a vector with ones (as 32bit integers) where the points are inliers
206 }
207 nr_p += _mm_extract_epi32 (res, 0);
208 nr_p += _mm_extract_epi32 (res, 1);
209 nr_p += _mm_extract_epi32 (res, 2);
210 nr_p += _mm_extract_epi32 (res, 3);
211
212 // Process the remaining points (at most 3)
213 nr_p += countWithinDistanceStandard(model_coefficients, threshold, i);
214 return (nr_p);
215}
216#endif
217
218//////////////////////////////////////////////////////////////////////////
219#if defined (__AVX__) && defined (__AVX2__)
220template <typename PointT, typename PointNT> std::size_t
222 const Eigen::VectorXf &model_coefficients, const double threshold, std::size_t i) const
223{
224 std::size_t nr_p = 0;
225 const __m256 a_vec = _mm256_set1_ps (model_coefficients[0]);
226 const __m256 b_vec = _mm256_set1_ps (model_coefficients[1]);
227 const __m256 c_vec = _mm256_set1_ps (model_coefficients[2]);
228 const __m256 d_vec = _mm256_set1_ps (model_coefficients[3]);
229 const __m256 threshold_vec = _mm256_set1_ps (threshold);
230 const __m256 normal_distance_weight_vec = _mm256_set1_ps (normal_distance_weight_);
231 const __m256 abs_help = _mm256_set1_ps (-0.0F); // -0.0F (negative zero) means that all bits are 0, only the sign bit is 1
232 __m256i res = _mm256_set1_epi32(0); // This corresponds to nr_p: 8 32bit integers that, summed together, hold the number of inliers
233 for (; (i + 8) <= indices_->size (); i += 8)
234 {
235 const __m256 d_euclid_vec = pcl::SampleConsensusModelPlane<PointT>::dist8 (i, a_vec, b_vec, c_vec, d_vec, abs_help);
236
237 const __m256 d_normal_vec = getAcuteAngle3DAVX (
238 _mm256_set_ps ((*normals_)[(*indices_)[i ]].normal_x,
239 (*normals_)[(*indices_)[i+1]].normal_x,
240 (*normals_)[(*indices_)[i+2]].normal_x,
241 (*normals_)[(*indices_)[i+3]].normal_x,
242 (*normals_)[(*indices_)[i+4]].normal_x,
243 (*normals_)[(*indices_)[i+5]].normal_x,
244 (*normals_)[(*indices_)[i+6]].normal_x,
245 (*normals_)[(*indices_)[i+7]].normal_x),
246 _mm256_set_ps ((*normals_)[(*indices_)[i ]].normal_y,
247 (*normals_)[(*indices_)[i+1]].normal_y,
248 (*normals_)[(*indices_)[i+2]].normal_y,
249 (*normals_)[(*indices_)[i+3]].normal_y,
250 (*normals_)[(*indices_)[i+4]].normal_y,
251 (*normals_)[(*indices_)[i+5]].normal_y,
252 (*normals_)[(*indices_)[i+6]].normal_y,
253 (*normals_)[(*indices_)[i+7]].normal_y),
254 _mm256_set_ps ((*normals_)[(*indices_)[i ]].normal_z,
255 (*normals_)[(*indices_)[i+1]].normal_z,
256 (*normals_)[(*indices_)[i+2]].normal_z,
257 (*normals_)[(*indices_)[i+3]].normal_z,
258 (*normals_)[(*indices_)[i+4]].normal_z,
259 (*normals_)[(*indices_)[i+5]].normal_z,
260 (*normals_)[(*indices_)[i+6]].normal_z,
261 (*normals_)[(*indices_)[i+7]].normal_z),
262 a_vec, b_vec, c_vec);
263 const __m256 weight_vec = _mm256_mul_ps (normal_distance_weight_vec, _mm256_sub_ps (_mm256_set1_ps (1.0f),
264 _mm256_set_ps ((*normals_)[(*indices_)[i ]].curvature,
265 (*normals_)[(*indices_)[i+1]].curvature,
266 (*normals_)[(*indices_)[i+2]].curvature,
267 (*normals_)[(*indices_)[i+3]].curvature,
268 (*normals_)[(*indices_)[i+4]].curvature,
269 (*normals_)[(*indices_)[i+5]].curvature,
270 (*normals_)[(*indices_)[i+6]].curvature,
271 (*normals_)[(*indices_)[i+7]].curvature)));
272 const __m256 dist = _mm256_andnot_ps (abs_help, _mm256_add_ps (_mm256_mul_ps (weight_vec, d_normal_vec), _mm256_mul_ps (_mm256_sub_ps (_mm256_set1_ps (1.0f), weight_vec), d_euclid_vec)));
273 const __m256 mask = _mm256_cmp_ps (dist, threshold_vec, _CMP_LT_OQ); // The mask contains 1 bits if the corresponding points are inliers, else 0 bits
274 res = _mm256_add_epi32 (res, _mm256_and_si256 (_mm256_set1_epi32 (1), _mm256_castps_si256 (mask))); // The latter part creates a vector with ones (as 32bit integers) where the points are inliers
275 }
276 nr_p += _mm256_extract_epi32 (res, 0);
277 nr_p += _mm256_extract_epi32 (res, 1);
278 nr_p += _mm256_extract_epi32 (res, 2);
279 nr_p += _mm256_extract_epi32 (res, 3);
280 nr_p += _mm256_extract_epi32 (res, 4);
281 nr_p += _mm256_extract_epi32 (res, 5);
282 nr_p += _mm256_extract_epi32 (res, 6);
283 nr_p += _mm256_extract_epi32 (res, 7);
284
285 // Process the remaining points (at most 7)
286 nr_p += countWithinDistanceStandard(model_coefficients, threshold, i);
287 return (nr_p);
288}
289#endif
290
291//////////////////////////////////////////////////////////////////////////////////////////////////////////////////
292template <typename PointT, typename PointNT> void
294 const Eigen::VectorXf &model_coefficients, std::vector<double> &distances) const
295{
296 if (!normals_)
297 {
298 PCL_ERROR ("[pcl::SampleConsensusModelNormalPlane::getDistancesToModel] No input dataset containing normals was given!\n");
299 return;
300 }
301
302 // Check if the model is valid given the user constraints
303 if (!isModelValid (model_coefficients))
304 {
305 distances.clear ();
306 return;
307 }
308
309 // Obtain the plane normal
310 Eigen::Vector4f coeff = model_coefficients;
311 coeff[3] = 0.0f;
312
313 distances.resize (indices_->size ());
314
315 // Iterate through the 3d points and calculate the distances from them to the plane
316 for (std::size_t i = 0; i < indices_->size (); ++i)
317 {
318 const PointT &pt = (*input_)[(*indices_)[i]];
319 const PointNT &nt = (*normals_)[(*indices_)[i]];
320 // Calculate the distance from the point to the plane normal as the dot product
321 // D = (P-A).N/|N|
322 Eigen::Vector4f p (pt.x, pt.y, pt.z, 0.0f);
323 Eigen::Vector4f n (nt.normal_x, nt.normal_y, nt.normal_z, 0.0f);
324 double d_euclid = std::abs (coeff.dot (p) + model_coefficients[3]);
325
326 // Calculate the angular distance between the point normal and the plane normal
327 double d_normal = std::abs (getAngle3D (n, coeff));
328 d_normal = (std::min) (d_normal, M_PI - d_normal);
329
330 // Weight with the point curvature. On flat surfaces, curvature -> 0, which means the normal will have a higher influence
331 double weight = normal_distance_weight_ * (1.0 - nt.curvature);
332
333 distances[i] = std::abs (weight * d_normal + (1.0 - weight) * d_euclid);
334 }
335}
336
337#define PCL_INSTANTIATE_SampleConsensusModelNormalPlane(PointT, PointNT) template class PCL_EXPORTS pcl::SampleConsensusModelNormalPlane<PointT, PointNT>;
338
339#endif // PCL_SAMPLE_CONSENSUS_IMPL_SAC_MODEL_NORMAL_PLANE_H_
340
SampleConsensusModelNormalPlane defines a model for 3D plane segmentation using additional surface no...
std::size_t countWithinDistance(const Eigen::VectorXf &model_coefficients, const double threshold) const override
Count all the points which respect the given model coefficients as inliers.
void selectWithinDistance(const Eigen::VectorXf &model_coefficients, const double threshold, Indices &inliers) override
Select all the points which respect the given model coefficients as inliers.
void getDistancesToModel(const Eigen::VectorXf &model_coefficients, std::vector< double > &distances) const override
Compute all distances from the cloud data to a given plane model.
std::size_t countWithinDistanceStandard(const Eigen::VectorXf &model_coefficients, const double threshold, std::size_t i=0) const
This implementation uses no SIMD instructions.
SampleConsensusModelPlane defines a model for 3D plane segmentation.
Define standard C methods and C++ classes that are common to all methods.
double getAngle3D(const Eigen::Vector4f &v1, const Eigen::Vector4f &v2, const bool in_degree=false)
Compute the smallest angle between two 3D vectors in radians (default) or degree.
Definition: common.hpp:47
float distance(const PointT &p1, const PointT &p2)
Definition: geometry.h:60
IndicesAllocator<> Indices
Type used for indices in PCL.
Definition: types.h:133
#define M_PI
Definition: pcl_macros.h:201
A point structure representing Euclidean xyz coordinates, and the RGB color.