Point Cloud Library (PCL)
1.3.1
|
00001 /* 00002 * Software License Agreement (BSD License) 00003 * 00004 * Point Cloud Library (PCL) - www.pointclouds.org 00005 * Copyright (c) 2010-2011, Willow Garage, Inc. 00006 * 00007 * All rights reserved. 00008 * 00009 * Redistribution and use in source and binary forms, with or without 00010 * modification, are permitted provided that the following conditions 00011 * are met: 00012 * 00013 * * Redistributions of source code must retain the above copyright 00014 * notice, this list of conditions and the following disclaimer. 00015 * * Redistributions in binary form must reproduce the above 00016 * copyright notice, this list of conditions and the following 00017 * disclaimer in the documentation and/or other materials provided 00018 * with the distribution. 00019 * * Neither the name of Willow Garage, Inc. nor the names of its 00020 * contributors may be used to endorse or promote products derived 00021 * from this software without specific prior written permission. 00022 * 00023 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 00024 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT 00025 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS 00026 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE 00027 * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, 00028 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, 00029 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; 00030 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER 00031 * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 00032 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN 00033 * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 00034 * POSSIBILITY OF SUCH DAMAGE. 00035 * 00036 * 00037 */ 00038 #ifndef PCL_REGISTRATION_IMPL_CORRESPONDENCE_ESTIMATION_H_ 00039 #define PCL_REGISTRATION_IMPL_CORRESPONDENCE_ESTIMATION_H_ 00040 00041 #include <pcl/common/concatenate.h> 00042 //#include <pcl/registration/correspondence_estimation.h> 00043 00045 template <typename PointSource, typename PointTarget> inline void 00046 pcl::registration::CorrespondenceEstimation<PointSource, PointTarget>::setInputTarget ( 00047 const PointCloudTargetConstPtr &cloud) 00048 { 00049 if (cloud->points.empty ()) 00050 { 00051 PCL_ERROR ("[pcl::%s::setInputTarget] Invalid or empty point cloud dataset given!\n", getClassName ().c_str ()); 00052 return; 00053 } 00054 target_ = cloud; 00055 tree_->setInputCloud (target_); 00056 } 00057 00059 template <typename PointSource, typename PointTarget> void 00060 pcl::registration::CorrespondenceEstimation<PointSource, PointTarget>::determineCorrespondences ( 00061 pcl::Correspondences &correspondences, float max_distance) 00062 { 00063 typedef typename pcl::traits::fieldList<PointTarget>::type FieldListTarget; 00064 00065 if (!initCompute ()) 00066 return; 00067 00068 if (!target_) 00069 { 00070 PCL_WARN ("[pcl::%s::compute] No input target dataset was given!\n", getClassName ().c_str ()); 00071 return; 00072 } 00073 00074 float max_dist_sqr = max_distance * max_distance; 00075 00076 correspondences.resize (indices_->size ()); 00077 std::vector<int> index (1); 00078 std::vector<float> distance (1); 00079 pcl::Correspondence corr; 00080 for (size_t i = 0; i < indices_->size (); ++i) 00081 { 00082 // Copy the source data to a target PointTarget format so we can search in the tree 00083 PointTarget pt; 00084 pcl::for_each_type <FieldListTarget> (pcl::NdConcatenateFunctor <PointSource, PointTarget> ( 00085 input_->points[(*indices_)[i]], 00086 pt)); 00087 00088 //if (tree_->nearestKSearch (input_->points[(*indices_)[i]], 1, index, distance)) 00089 if (tree_->nearestKSearch (pt, 1, index, distance)) 00090 { 00091 if (distance[0] <= max_dist_sqr) 00092 { 00093 corr.index_query = i; 00094 corr.index_match = index[0]; 00095 corr.distance = distance[0]; 00096 correspondences[i] = corr; 00097 continue; 00098 } 00099 } 00100 // correspondences[i] = pcl::Correspondence(i, -1, std::numeric_limits<float>::max()); 00101 } 00102 deinitCompute (); 00103 } 00104 00106 template <typename PointSource, typename PointTarget> void 00107 pcl::registration::CorrespondenceEstimation<PointSource, PointTarget>::determineReciprocalCorrespondences ( 00108 pcl::Correspondences &correspondences) 00109 { 00110 typedef typename pcl::traits::fieldList<PointSource>::type FieldListSource; 00111 typedef typename pcl::traits::fieldList<PointTarget>::type FieldListTarget; 00112 typedef typename pcl::intersect<FieldListSource, FieldListTarget>::type FieldList; 00113 00114 if (!initCompute ()) 00115 return; 00116 00117 if (!target_) 00118 { 00119 PCL_WARN ("[pcl::%s::compute] No input target dataset was given!\n", getClassName ().c_str ()); 00120 return; 00121 } 00122 00123 // setup tree for reciprocal search 00124 pcl::KdTreeFLANN<PointSource> tree_reciprocal; 00125 tree_reciprocal.setInputCloud (input_, indices_); 00126 00127 correspondences.resize (indices_->size()); 00128 std::vector<int> index (1); 00129 std::vector<float> distance (1); 00130 std::vector<int> index_reciprocal (1); 00131 std::vector<float> distance_reciprocal (1); 00132 pcl::Correspondence corr; 00133 unsigned int nr_valid_correspondences = 0; 00134 00135 for (size_t i = 0; i < indices_->size (); ++i) 00136 { 00137 // Copy the source data to a target PointTarget format so we can search in the tree 00138 PointTarget pt_src; 00139 pcl::for_each_type <FieldList> (pcl::NdConcatenateFunctor <PointSource, PointTarget> ( 00140 input_->points[(*indices_)[i]], 00141 pt_src)); 00142 00143 //tree_->nearestKSearch (input_->points[(*indices_)[i]], 1, index, distance); 00144 tree_->nearestKSearch (pt_src, 1, index, distance); 00145 00146 // Copy the target data to a target PointSource format so we can search in the tree_reciprocal 00147 PointSource pt_tgt; 00148 pcl::for_each_type <FieldList> (pcl::NdConcatenateFunctor <PointTarget, PointSource> ( 00149 target_->points[index[0]], 00150 pt_tgt)); 00151 //tree_reciprocal.nearestKSearch (target_->points[index[0]], 1, index_reciprocal, distance_reciprocal); 00152 tree_reciprocal.nearestKSearch (pt_tgt, 1, index_reciprocal, distance_reciprocal); 00153 00154 if ((*indices_)[i] == index_reciprocal[0]) 00155 { 00156 corr.index_query = (*indices_)[i]; 00157 corr.index_match = index[0]; 00158 corr.distance = distance[0]; 00159 correspondences[nr_valid_correspondences] = corr; 00160 ++nr_valid_correspondences; 00161 } 00162 } 00163 correspondences.resize (nr_valid_correspondences); 00164 00165 deinitCompute (); 00166 } 00167 00168 //#define PCL_INSTANTIATE_CorrespondenceEstimation(T,U) template class PCL_EXPORTS pcl::registration::CorrespondenceEstimation<T,U>; 00169 00170 #endif /* PCL_REGISTRATION_IMPL_CORRESPONDENCE_ESTIMATION_H_ */