1 /*
2  * Software License Agreement (BSD License)
3  *
4  * Point Cloud Library (PCL) - www.pointclouds.org
5  * Copyright (c) 2013-, Open Perception, 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 the copyright holder(s) 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  * ground_based_people_detection_app.h
37  * Created on: Nov 30, 2012
38  * Author: Matteo Munaro
39  */
40 
41 #pragma once
42 
43 #include <pcl/point_types.h>
44 #include <pcl/sample_consensus/sac_model_plane.h>
45 #include <pcl/sample_consensus/ransac.h>
46 #include <pcl/kdtree/kdtree.h>
47 #include <pcl/people/person_cluster.h>
48 #include <pcl/people/head_based_subcluster.h>
49 #include <pcl/people/person_classifier.h>
50 #include <pcl/common/transforms.h>
51 
52 namespace pcl
53 {
54   namespace people
55   {
56     /** \brief GroundBasedPeopleDetectionApp performs people detection on RGB-D data having as input the ground plane coefficients.
57      * It implements the people detection algorithm described here:
58      * M. Munaro, F. Basso and E. Menegatti,
59      * Tracking people within groups with RGB-D data,
60      * In Proceedings of the International Conference on Intelligent Robots and Systems (IROS) 2012, Vilamoura (Portugal), 2012.
61      *
62      * \author Matteo Munaro
63      * \ingroup people
64      */
65     template <typename PointT> class GroundBasedPeopleDetectionApp;
66 
67     template <typename PointT>
68     class GroundBasedPeopleDetectionApp
69     {
70     public:
71 
72       using PointCloud = pcl::PointCloud<PointT>;
73       using PointCloudPtr = typename PointCloud::Ptr;
74       using PointCloudConstPtr = typename PointCloud::ConstPtr;
75 
76       /** \brief Constructor. */
77       GroundBasedPeopleDetectionApp ();
78 
79       /** \brief Destructor. */
80       virtual ~GroundBasedPeopleDetectionApp ();
81 
82       /**
83        * \brief Set the pointer to the input cloud.
84        *
85        * \param[in] cloud A pointer to the input cloud.
86        */
87       void
88       setInputCloud (PointCloudPtr& cloud);
89 
90       /**
91        * \brief Set the ground coefficients.
92        *
93        * \param[in] ground_coeffs Vector containing the four plane coefficients.
94        */
95       void
96       setGround (Eigen::VectorXf& ground_coeffs);
97 
98       /**
99        * \brief Set the transformation matrix, which is used in order to transform the given point cloud, the ground plane and the intrinsics matrix to the internal coordinate frame.
100        * \param[in] transformation
101        */
102       void
103       setTransformation (const Eigen::Matrix3f& transformation);
104 
105       /**
106        * \brief Set sampling factor.
107        *
108        * \param[in] sampling_factor Value of the downsampling factor (in each dimension) which is applied to the raw point cloud (default = 1.).
109        */
110       void
111       setSamplingFactor (int sampling_factor);
112 
113       /**
114        * \brief Set voxel size.
115        *
116        * \param[in] voxel_size Value of the voxel dimension (default = 0.06m.).
117        */
118       void
119       setVoxelSize (float voxel_size);
120 
121       /**
122        * \brief Set intrinsic parameters of the RGB camera.
123        *
124        * \param[in] intrinsics_matrix RGB camera intrinsic parameters matrix.
125        */
126       void
127       setIntrinsics (Eigen::Matrix3f intrinsics_matrix);
128 
129       /**
130        * \brief Set SVM-based person classifier.
131        *
132        * \param[in] person_classifier Needed for people detection on RGB data.
133        */
134       void
135       setClassifier (pcl::people::PersonClassifier<pcl::RGB> person_classifier);
136 
137       /**
138        * \brief Set the field of view of the point cloud in z direction.
139        *
140        * \param[in] min The beginning of the field of view in z-direction, should be usually set to zero.
141        * \param[in] max The end of the field of view in z-direction.
142        */
143       void
144       setFOV (float min, float max);
145 
146       /**
147        * \brief Set sensor orientation (vertical = true means portrait mode, vertical = false means landscape mode).
148        *
149        * \param[in] vertical Set landscape/portrait camera orientation (default = false).
150        */
151       void
152       setSensorPortraitOrientation (bool vertical);
153 
154       /**
155        * \brief Set head_centroid_ to true (person centroid is in the head) or false (person centroid is the whole body centroid).
156        *
157        * \param[in] head_centroid Set the location of the person centroid (head or body center) (default = true).
158        */
159       void
160       setHeadCentroid (bool head_centroid);
161 
162       /**
163        * \brief Set minimum and maximum allowed height and width for a person cluster.
164        *
165        * \param[in] min_height Minimum allowed height for a person cluster (default = 1.3).
166        * \param[in] max_height Maximum allowed height for a person cluster (default = 2.3).
167        * \param[in] min_width Minimum width for a person cluster (default = 0.1).
168        * \param[in] max_width Maximum width for a person cluster (default = 8.0).
169        */
170       void
171       setPersonClusterLimits (float min_height, float max_height, float min_width, float max_width);
172 
173       /**
174        * \brief Set minimum distance between persons' heads.
175        *
176        * \param[in] heads_minimum_distance Minimum allowed distance between persons' heads (default = 0.3).
177        */
178       void
179       setMinimumDistanceBetweenHeads (float heads_minimum_distance);
180 
181       /**
182        * \brief Get the minimum and maximum allowed height and width for a person cluster.
183        *
184        * \param[out] min_height Minimum allowed height for a person cluster.
185        * \param[out] max_height Maximum allowed height for a person cluster.
186        * \param[out] min_width Minimum width for a person cluster.
187        * \param[out] max_width Maximum width for a person cluster.
188        */
189       void
190       getPersonClusterLimits (float& min_height, float& max_height, float& min_width, float& max_width);
191 
192       /**
193        * \brief Get minimum and maximum allowed number of points for a person cluster.
194        *
195        * \param[out] min_points Minimum allowed number of points for a person cluster.
196        * \param[out] max_points Maximum allowed number of points for a person cluster.
197        */
198       void
199       getDimensionLimits (int& min_points, int& max_points);
200 
201       /**
202        * \brief Get minimum distance between persons' heads.
203        */
204       float
205       getMinimumDistanceBetweenHeads ();
206 
207       /**
208        * \brief Get floor coefficients.
209        */
210       Eigen::VectorXf
211       getGround ();
212 
213       /**
214        * \brief Get the filtered point cloud.
215        */
216       PointCloudPtr
217       getFilteredCloud ();
218 
219       /**
220        * \brief Get pointcloud after voxel grid filtering and ground removal.
221        */
222       PointCloudPtr
223       getNoGroundCloud ();
224 
225       /**
226        * \brief Extract RGB information from a point cloud and output the corresponding RGB point cloud.
227        *
228        * \param[in] input_cloud A pointer to a point cloud containing also RGB information.
229        * \param[out] output_cloud A pointer to a RGB point cloud.
230        */
231       void
232       extractRGBFromPointCloud (PointCloudPtr input_cloud, pcl::PointCloud<pcl::RGB>::Ptr& output_cloud);
233 
234       /**
235        * \brief Swap rows/cols dimensions of a RGB point cloud (90 degrees counterclockwise rotation).
236        *
237        * \param[in,out] cloud A pointer to a RGB point cloud.
238        */
239       void
240       swapDimensions (pcl::PointCloud<pcl::RGB>::Ptr& cloud);
241 
242      /**
243        * \brief Estimates min_points_ and max_points_ based on the minimal and maximal cluster size and the voxel size.
244        */
245       void
246       updateMinMaxPoints ();
247 
248       /**
249        * \brief Applies the transformation to the input point cloud.
250        */
251       void
252       applyTransformationPointCloud ();
253 
254       /**
255        * \brief Applies the transformation to the ground plane.
256        */
257       void
258       applyTransformationGround ();
259 
260       /**
261        * \brief Applies the transformation to the intrinsics matrix.
262        */
263       void
264       applyTransformationIntrinsics ();
265 
266       /**
267        * \brief Reduces the input cloud to one point per voxel and limits the field of view.
268        */
269       void
270       filter ();
271 
272       /**
273        * \brief Perform people detection on the input data and return people clusters information.
274        *
275        * \param[out] clusters Vector of PersonCluster.
276        *
277        * \return true if the compute operation is successful, false otherwise.
278        */
279       bool
280       compute (std::vector<pcl::people::PersonCluster<PointT> >& clusters);
281 
282     protected:
283       /** \brief sampling factor used to downsample the point cloud */
284       int sampling_factor_;
285 
286       /** \brief voxel size */
287       float voxel_size_;
288 
289       /** \brief ground plane coefficients */
290       Eigen::VectorXf ground_coeffs_;
291 
292       /** \brief flag stating whether the ground coefficients have been set or not */
293       bool ground_coeffs_set_;
294 
295       /** \brief the transformed ground coefficients */
296       Eigen::VectorXf ground_coeffs_transformed_;
297 
298       /** \brief ground plane normalization factor */
299       float sqrt_ground_coeffs_;
300 
301       /** \brief rotation matrix which transforms input point cloud to internal people tracker coordinate frame */
302       Eigen::Matrix3f transformation_;
303 
304       /** \brief flag stating whether the transformation matrix has been set or not */
305       bool transformation_set_;
306 
307       /** \brief pointer to the input cloud */
308       PointCloudPtr cloud_;
309 
310       /** \brief pointer to the filtered cloud */
311       PointCloudPtr cloud_filtered_;
312 
313       /** \brief pointer to the cloud after voxel grid filtering and ground removal */
314       PointCloudPtr no_ground_cloud_;
315 
316       /** \brief pointer to a RGB cloud corresponding to cloud_ */
317       pcl::PointCloud<pcl::RGB>::Ptr rgb_image_;
318 
319       /** \brief person clusters maximum height from the ground plane */
320       float max_height_;
321 
322       /** \brief person clusters minimum height from the ground plane */
323       float min_height_;
324 
325       /** \brief person clusters maximum width, used to estimate how many points maximally represent a person cluster */
326       float max_width_;
327 
328       /** \brief person clusters minimum width, used to estimate how many points minimally represent a person cluster */
329       float min_width_;
330 
331       /** \brief the beginning of the field of view in z-direction, should be usually set to zero */
332       float min_fov_;
333 
334       /** \brief the end of the field of view in z-direction */
335       float max_fov_;
336 
337       /** \brief if true, the sensor is considered to be vertically placed (portrait mode) */
338       bool vertical_;
339 
340       /** \brief if true, the person centroid is computed as the centroid of the cluster points belonging to the head;
341        * if false, the person centroid is computed as the centroid of the whole cluster points (default = true) */
342       bool head_centroid_;    // if true, the person centroid is computed as the centroid of the cluster points belonging to the head (default = true)
343                               // if false, the person centroid is computed as the centroid of the whole cluster points
344       /** \brief maximum number of points for a person cluster */
345       int max_points_;
346 
347       /** \brief minimum number of points for a person cluster */
348       int min_points_;
349 
350       /** \brief minimum distance between persons' heads */
351       float heads_minimum_distance_;
352 
353       /** \brief intrinsic parameters matrix of the RGB camera */
354       Eigen::Matrix3f intrinsics_matrix_;
355 
356       /** \brief flag stating whether the intrinsics matrix has been set or not */
357       bool intrinsics_matrix_set_;
358 
359       /** \brief the transformed intrinsics matrix */
360       Eigen::Matrix3f intrinsics_matrix_transformed_;
361 
362       /** \brief SVM-based person classifier */
363       pcl::people::PersonClassifier<pcl::RGB> person_classifier_;
364 
365       /** \brief flag stating if the classifier has been set or not */
366       bool person_classifier_set_flag_;
367     };
368   } /* namespace people */
369 } /* namespace pcl */
370 #include <pcl/people/impl/ground_based_people_detection_app.hpp>
371