Point Cloud Library (PCL)  1.9.1-dev
standalone_marching_cubes.hpp
1  /*
2  * Software License Agreement (BSD License)
3  *
4  * Point Cloud Library (PCL) - www.pointclouds.org
5  * Copyright (c) 2011, Willow Garage, 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 Willow Garage, Inc. 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  */
37 
38 #ifndef PCL_STANDALONE_MARCHING_CUBES_IMPL_HPP_
39 #define PCL_STANDALONE_MARCHING_CUBES_IMPL_HPP_
40 
41 #include <pcl/gpu/kinfu_large_scale/standalone_marching_cubes.h>
42 
43 ///////////////////////////////////////////////////////////////////////////////
44 template <typename PointT>
45 pcl::gpu::kinfuLS::StandaloneMarchingCubes<PointT>::StandaloneMarchingCubes (int new_voxels_x, int new_voxels_y, int new_voxels_z, float new_volume_size)
46 {
47  voxels_x_ = new_voxels_x;
48  voxels_y_ = new_voxels_y;
49  voxels_z_ = new_voxels_z;
50  volume_size_ = new_volume_size;
51 
52  ///Creating GPU TSDF Volume instance
53  const Eigen::Vector3f volume_size = Eigen::Vector3f::Constant (volume_size_);
54  // std::cout << "VOLUME SIZE IS " << volume_size_ << std::endl;
55  const Eigen::Vector3i volume_resolution (voxels_x_, voxels_y_, voxels_z_);
56  tsdf_volume_gpu_ = TsdfVolume::Ptr ( new TsdfVolume (volume_resolution) );
57  tsdf_volume_gpu_->setSize (volume_size);
58 
59  ///Creating CPU TSDF Volume instance
60  int tsdf_total_size = voxels_x_ * voxels_y_ * voxels_z_;
61  tsdf_volume_cpu_= std::vector<int> (tsdf_total_size,0);
62 
63  mesh_counter_ = 0;
64 }
65 
66 ///////////////////////////////////////////////////////////////////////////////
67 
68 template <typename PointT> typename pcl::gpu::kinfuLS::StandaloneMarchingCubes<PointT>::MeshPtr
70 {
71 
72  //Clearing TSDF GPU and cPU
73  std::cout << "VOLUME SIZE IS " << volume_size_ << std::endl;
74 
75  //Clear values in TSDF Volume GPU
76  tsdf_volume_gpu_->reset (); // This one uses the same tsdf volume but clears it before loading new values. This one is our friend.
77 
78  //Clear values in TSDF Volume CPU
79  fill (tsdf_volume_cpu_.begin (), tsdf_volume_cpu_.end (), 0);
80 
81  //Loading values to GPU
82  loadTsdfCloudToGPU (cloud);
83 
84  //Creating and returning mesh
85  return ( runMarchingCubes () );
86 
87 }
88 
89 ///////////////////////////////////////////////////////////////////////////////
90 
91 //template <typename PointT> std::vector< typename pcl::gpu::StandaloneMarchingCubes<PointT>::MeshPtr >
92 template <typename PointT> void
93 pcl::gpu::kinfuLS::StandaloneMarchingCubes<PointT>::getMeshesFromTSDFVector (const std::vector<PointCloudPtr> &tsdf_clouds, const std::vector<Eigen::Vector3f, Eigen::aligned_allocator<Eigen::Vector3f> > &tsdf_offsets)
94 {
95  std::vector< MeshPtr > meshes_vector;
96 
97  int max_iterations = std::min( tsdf_clouds.size (), tsdf_offsets.size () ); //Safety check
98  PCL_INFO ("There are %d cubes to be processed \n", max_iterations);
99  float cell_size = volume_size_ / voxels_x_;
100 
101  int mesh_counter = 0;
102 
103  for(int i = 0; i < max_iterations; ++i)
104  {
105  PCL_INFO ("Processing cube number %d\n", i);
106 
107  //Making cloud local
108  Eigen::Affine3f cloud_transform;
109 
110  float originX = (tsdf_offsets[i]).x();
111  float originY = (tsdf_offsets[i]).y();
112  float originZ = (tsdf_offsets[i]).z();
113 
114  cloud_transform.linear ().setIdentity ();
115  cloud_transform.translation ()[0] = -originX;
116  cloud_transform.translation ()[1] = -originY;
117  cloud_transform.translation ()[2] = -originZ;
118 
119  transformPointCloud (*tsdf_clouds[i], *tsdf_clouds[i], cloud_transform);
120 
121  //Get mesh
122  MeshPtr tmp = getMeshFromTSDFCloud (*tsdf_clouds[i]);
123 
124  if(tmp != nullptr)
125  {
126  meshes_vector.push_back (tmp);
127  mesh_counter++;
128  }
129  else
130  {
131  PCL_INFO ("This cloud returned no faces, we skip it!\n");
132  continue;
133  }
134 
135  //Making cloud global
136  cloud_transform.translation ()[0] = originX * cell_size;
137  cloud_transform.translation ()[1] = originY * cell_size;
138  cloud_transform.translation ()[2] = originZ * cell_size;
139 
141  fromPCLPointCloud2 ( (meshes_vector.back () )->cloud, *cloud_tmp_ptr);
142 
143  transformPointCloud (*cloud_tmp_ptr, *cloud_tmp_ptr, cloud_transform);
144 
145  toPCLPointCloud2 (*cloud_tmp_ptr, (meshes_vector.back () )->cloud);
146 
147  std::stringstream name;
148  name << "mesh_" << mesh_counter << ".ply";
149  PCL_INFO ("Saving mesh...%d \n", mesh_counter);
150  pcl::io::savePLYFile (name.str (), *(meshes_vector.back ()));
151 
152  }
153  return;
154 }
155 
156 ///////////////////////////////////////////////////////////////////////////////
157 
158 template <typename PointT> pcl::gpu::kinfuLS::TsdfVolume::Ptr
160 {
161  return (tsdf_volume_gpu_);
162 }
163 
164 ///////////////////////////////////////////////////////////////////////////////
165 
166 template <typename PointT> std::vector<int>& //todo
168 {
169  return (tsdf_volume_cpu_);
170 }
171 
172 ///////////////////////////////////////////////////////////////////////////////
173 
174 template <typename PointT> void
176 {
177  //Converting Values
178  convertTsdfVectors (cloud, tsdf_volume_cpu_);
179 
180  //Uploading data to GPU
181  int cubeColumns = voxels_x_;
182  tsdf_volume_gpu_->data ().upload (tsdf_volume_cpu_, cubeColumns);
183 }
184 
185 ///////////////////////////////////////////////////////////////////////////////
186 
187 template <typename PointT> void
189 {
190  constexpr int DIVISOR = std::numeric_limits<short>::max();
191 
192  ///For every point in the cloud
193 #pragma omp parallel for
194 
195  for(int i = 0; i < (int) cloud.points.size (); ++i)
196  {
197  int x = cloud.points[i].x;
198  int y = cloud.points[i].y;
199  int z = cloud.points[i].z;
200 
201  if(x > 0 && x < voxels_x_ && y > 0 && y < voxels_y_ && z > 0 && z < voxels_z_)
202  {
203  ///Calculate the index to write to
204  int dst_index = x + voxels_x_ * y + voxels_y_ * voxels_x_ * z;
205 
206  short2& elem = *reinterpret_cast<short2*> (&output[dst_index]);
207  elem.x = static_cast<short> (cloud.points[i].intensity * DIVISOR);
208  elem.y = static_cast<short> (1);
209  }
210  }
211 }
212 
213 ///////////////////////////////////////////////////////////////////////////////
214 
215 template <typename PointT> typename pcl::gpu::kinfuLS::StandaloneMarchingCubes<PointT>::MeshPtr
217 {
218  if (triangles.empty () )
219  {
220  return MeshPtr ();
221  }
222 
224  cloud.width = (int)triangles.size ();
225  cloud.height = 1;
226  triangles.download (cloud.points);
227 
228  boost::shared_ptr<pcl::PolygonMesh> mesh_ptr ( new pcl::PolygonMesh () );
229 
230  pcl::toPCLPointCloud2 (cloud, mesh_ptr->cloud);
231 
232  mesh_ptr->polygons.resize (triangles.size () / 3);
233  for (std::size_t i = 0; i < mesh_ptr->polygons.size (); ++i)
234  {
235  pcl::Vertices v;
236  v.vertices.push_back (i*3+0);
237  v.vertices.push_back (i*3+2);
238  v.vertices.push_back (i*3+1);
239  mesh_ptr->polygons[i] = v;
240  }
241  return (mesh_ptr);
242 }
243 
244 ///////////////////////////////////////////////////////////////////////////////
245 
246 template <typename PointT> typename pcl::gpu::kinfuLS::StandaloneMarchingCubes<PointT>::MeshPtr
248 {
249  //Preparing the pointers and variables
250  const TsdfVolume::Ptr tsdf_volume_const_ = tsdf_volume_gpu_;
251  pcl::gpu::DeviceArray<pcl::PointXYZ> triangles_buffer_device_;
252 
253  //Creating Marching cubes instance
254  MarchingCubes::Ptr marching_cubes_ = MarchingCubes::Ptr ( new MarchingCubes() );
255 
256  //Running marching cubes
257  pcl::gpu::DeviceArray<pcl::PointXYZ> triangles_device = marching_cubes_->run (*tsdf_volume_const_, triangles_buffer_device_);
258 
259  //Creating mesh
260  boost::shared_ptr<pcl::PolygonMesh> mesh_ptr_ = convertTrianglesToMesh (triangles_device);
261 
262  if(mesh_ptr_ != nullptr)
263  {
265  fromPCLPointCloud2 ( mesh_ptr_->cloud, *cloud_tmp_ptr);
266  }
267  return (mesh_ptr_);
268 }
269 
270 ///////////////////////////////////////////////////////////////////////////////
271 
272 #endif // PCL_STANDALONE_MARCHING_CUBES_IMPL_HPP_
273 
void convertTsdfVectors(const PointCloud &cloud, std::vector< int > &output)
Read the data in the point cloud.
void fromPCLPointCloud2(const pcl::PCLPointCloud2 &msg, pcl::PointCloud< PointT > &cloud, const MsgFieldMap &field_map)
Convert a PCLPointCloud2 binary data blob into a pcl::PointCloud<T> object using a field_map...
Definition: conversions.h:168
bool empty() const
Returns true if unallocated otherwise false.
std::vector< std::uint32_t > vertices
Definition: Vertices.h:19
std::size_t size() const
Returns size in elements.
std::vector< PointT, Eigen::aligned_allocator< PointT > > points
The point data.
Definition: point_cloud.h:394
boost::shared_ptr< TsdfVolume > Ptr
Definition: tsdf_volume.h:64
int savePLYFile(const std::string &file_name, const pcl::PCLPointCloud2 &cloud, const Eigen::Vector4f &origin=Eigen::Vector4f::Zero(), const Eigen::Quaternionf &orientation=Eigen::Quaternionf::Identity(), bool binary_mode=false, bool use_camera=true)
Save point cloud data to a PLY file containing n-D points.
Definition: ply_io.h:810
std::vector< int > & tsdfVolumeCPU()
Returns the associated Tsdf Volume buffer in CPU.
void getMeshesFromTSDFVector(const std::vector< PointCloudPtr > &tsdf_clouds, const std::vector< Eigen::Vector3f, Eigen::aligned_allocator< Eigen::Vector3f > > &tsdf_offsets)
Runs marching cubes on every pointcloud in the vector.
boost::shared_ptr< pcl::PolygonMesh > MeshPtr
Describes a set of vertices in a polygon mesh, by basically storing an array of indices.
Definition: Vertices.h:14
MarchingCubes implements MarchingCubes functionality for TSDF volume on GPU.
std::uint32_t width
The point cloud width (if organized as an image-structure).
Definition: point_cloud.h:397
boost::shared_ptr< MarchingCubes > Ptr
Smart pointer.
void transformPointCloud(const pcl::PointCloud< PointT > &cloud_in, pcl::PointCloud< PointT > &cloud_out, const Eigen::Transform< Scalar, 3, Eigen::Affine > &transform, bool copy_all_fields=true)
Apply an affine transform defined by an Eigen Transform.
Definition: transforms.hpp:215
TsdfVolume class.
Definition: tsdf_volume.h:61
boost::shared_ptr< PointCloud< PointT > > Ptr
Definition: point_cloud.h:412
std::uint32_t height
The point cloud height (if organized as an image-structure).
Definition: point_cloud.h:399
void download(T *host_ptr) const
Downloads data from internal buffer to CPU memory.
PointCloud represents the base class in PCL for storing collections of 3D points. ...
MeshPtr getMeshFromTSDFCloud(const PointCloud &cloud)
Run marching cubes in a TSDF cloud and returns a PolygonMesh.
void toPCLPointCloud2(const pcl::PointCloud< PointT > &cloud, pcl::PCLPointCloud2 &msg)
Convert a pcl::PointCloud<T> object to a PCLPointCloud2 binary data blob.
Definition: conversions.h:241
MeshPtr convertTrianglesToMesh(const pcl::gpu::DeviceArray< pcl::PointXYZ > &triangles)
Converts the triangles buffer device to a PolygonMesh.
TsdfVolume::Ptr tsdfVolumeGPU()
Returns the associated Tsdf Volume buffer in GPU.
MeshPtr runMarchingCubes()
Runs marching cubes on the data that is contained in the TSDF Volume in GPU.
void loadTsdfCloudToGPU(const PointCloud &cloud)
Loads a TSDF Cloud to the TSDF Volume in GPU.
StandaloneMarchingCubes(int voxels_x=512, int voxels_y=512, int voxels_z=512, float volume_size=3.0f)
Constructor.