Go to the documentation of this file.00001
00002
00003
00004
00005
00006
00007
00008
00009
00010
00011
00012
00013
00014
00015
00016
00017
00018
00019
00020
00021
00022
00023
00024
00025
00026
00027
00028
00029
00030
00031
00032
00033
00034
00035
00036
00037
00038
00039 #ifndef PCL_SURFACE_IMPL_MARCHING_CUBES_RBF_H_
00040 #define PCL_SURFACE_IMPL_MARCHING_CUBES_RBF_H_
00041
00042 #include <pcl/surface/marching_cubes_rbf.h>
00043 #include <pcl/common/common.h>
00044 #include <pcl/common/vector_average.h>
00045 #include <pcl/Vertices.h>
00046 #include <pcl/kdtree/kdtree_flann.h>
00047
00049 template <typename PointNT>
00050 pcl::MarchingCubesRBF<PointNT>::MarchingCubesRBF ()
00051 : MarchingCubes<PointNT> (),
00052 off_surface_epsilon_ (0.1f)
00053 {
00054 }
00055
00057 template <typename PointNT>
00058 pcl::MarchingCubesRBF<PointNT>::~MarchingCubesRBF ()
00059 {
00060 }
00061
00063 template <typename PointNT> void
00064 pcl::MarchingCubesRBF<PointNT>::voxelizeData ()
00065 {
00066
00067 unsigned int N = static_cast<unsigned int> (input_->size ());
00068 Eigen::MatrixXd M (2*N, 2*N),
00069 d (2*N, 1);
00070
00071 for (unsigned int row_i = 0; row_i < 2*N; ++row_i)
00072 {
00073
00074 bool row_off = (row_i >= N) ? 1 : 0;
00075 for (unsigned int col_i = 0; col_i < 2*N; ++col_i)
00076 {
00077
00078 bool col_off = (col_i >= N) ? 1 : 0;
00079 M (row_i, col_i) = kernel (Eigen::Vector3f (input_->points[col_i%N].getVector3fMap ()).cast<double> () + Eigen::Vector3f (input_->points[col_i%N].getNormalVector3fMap ()).cast<double> () * col_off * off_surface_epsilon_,
00080 Eigen::Vector3f (input_->points[row_i%N].getVector3fMap ()).cast<double> () + Eigen::Vector3f (input_->points[row_i%N].getNormalVector3fMap ()).cast<double> () * row_off * off_surface_epsilon_);
00081 }
00082
00083 d (row_i, 0) = row_off * off_surface_epsilon_;
00084 }
00085
00086
00087 Eigen::MatrixXd w (2*N, 1);
00088
00089
00090 w = M.fullPivLu ().solve (d);
00091
00092 std::vector<double> weights (2*N);
00093 std::vector<Eigen::Vector3d> centers (2*N);
00094 for (unsigned int i = 0; i < N; ++i)
00095 {
00096 centers[i] = Eigen::Vector3f (input_->points[i].getVector3fMap ()).cast<double> ();
00097 centers[i + N] = Eigen::Vector3f (input_->points[i].getVector3fMap ()).cast<double> () + Eigen::Vector3f (input_->points[i].getNormalVector3fMap ()).cast<double> () * off_surface_epsilon_;
00098 weights[i] = w (i, 0);
00099 weights[i + N] = w (i + N, 0);
00100 }
00101
00102 for (int x = 0; x < res_x_; ++x)
00103 for (int y = 0; y < res_y_; ++y)
00104 for (int z = 0; z < res_z_; ++z)
00105 {
00106 Eigen::Vector3d point;
00107 point[0] = min_p_[0] + (max_p_[0] - min_p_[0]) * float (x) / float (res_x_);
00108 point[1] = min_p_[1] + (max_p_[1] - min_p_[1]) * float (y) / float (res_y_);
00109 point[2] = min_p_[2] + (max_p_[2] - min_p_[2]) * float (z) / float (res_z_);
00110
00111 double f = 0.0;
00112 std::vector<double>::const_iterator w_it (weights.begin());
00113 for (std::vector<Eigen::Vector3d>::const_iterator c_it = centers.begin ();
00114 c_it != centers.end (); ++c_it, ++w_it)
00115 f += *w_it * kernel (*c_it, point);
00116
00117 grid_[x * res_y_*res_z_ + y * res_z_ + z] = float (f);
00118 }
00119 }
00120
00122 template <typename PointNT> double
00123 pcl::MarchingCubesRBF<PointNT>::kernel (Eigen::Vector3d c, Eigen::Vector3d x)
00124 {
00125 double r = (x - c).norm ();
00126 return (r * r * r);
00127 }
00128
00129 #define PCL_INSTANTIATE_MarchingCubesRBF(T) template class PCL_EXPORTS pcl::MarchingCubesRBF<T>;
00130
00131 #endif // PCL_SURFACE_IMPL_MARCHING_CUBES_HOPPE_H_
00132