Point Cloud Library (PCL)  1.11.0
fast_bilateral_omp.hpp
1 /*
2  * Software License Agreement (BSD License)
3  *
4  * Point Cloud Library (PCL) - www.pointclouds.org
5  * Copyright (c) 2012-, Open Perception, Inc.
6  * Copyright (c) 2004, Sylvain Paris and Francois Sillion
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: fast_bilateral_omp.hpp 8381 2013-01-02 23:12:44Z sdmiller $
38  *
39  */
40 #ifndef PCL_FILTERS_IMPL_FAST_BILATERAL_OMP_HPP_
41 #define PCL_FILTERS_IMPL_FAST_BILATERAL_OMP_HPP_
42 
43 #include <pcl/filters/fast_bilateral_omp.h>
44 #include <pcl/common/io.h>
45 #include <cassert>
46 
47 //////////////////////////////////////////////////////////////////////////////////////////////
48 template <typename PointT> void
50 {
51  if (nr_threads == 0)
52 #ifdef _OPENMP
53  threads_ = omp_get_num_procs();
54 #else
55  threads_ = 1;
56 #endif
57  else
58  threads_ = nr_threads;
59 }
60 
61 //////////////////////////////////////////////////////////////////////////////////////////////
62 template <typename PointT> void
64 {
65  if (!input_->isOrganized ())
66  {
67  PCL_ERROR ("[pcl::FastBilateralFilterOMP] Input cloud needs to be organized.\n");
68  return;
69  }
70 
71  copyPointCloud (*input_, output);
72  float base_max = -std::numeric_limits<float>::max (),
73  base_min = std::numeric_limits<float>::max ();
74  bool found_finite = false;
75  for (std::size_t x = 0; x < output.width; ++x)
76  {
77  for (std::size_t y = 0; y < output.height; ++y)
78  {
79  if (std::isfinite (output (x, y).z))
80  {
81  if (base_max < output (x, y).z)
82  base_max = output (x, y).z;
83  if (base_min > output (x, y).z)
84  base_min = output (x, y).z;
85  found_finite = true;
86  }
87  }
88  }
89  if (!found_finite)
90  {
91  PCL_WARN ("[pcl::FastBilateralFilterOMP] Given an empty cloud. Doing nothing.\n");
92  return;
93  }
94 #pragma omp parallel for \
95  default(none) \
96  shared(base_min, base_max, output) \
97  num_threads(threads_)
98  for (long int i = 0; i < static_cast<long int> (output.size ()); ++i)
99  if (!std::isfinite (output.at(i).z))
100  output.at(i).z = base_max;
101 
102  const float base_delta = base_max - base_min;
103 
104  const std::size_t padding_xy = 2;
105  const std::size_t padding_z = 2;
106 
107  const std::size_t small_width = static_cast<std::size_t> (static_cast<float> (input_->width - 1) / sigma_s_) + 1 + 2 * padding_xy;
108  const std::size_t small_height = static_cast<std::size_t> (static_cast<float> (input_->height - 1) / sigma_s_) + 1 + 2 * padding_xy;
109  const std::size_t small_depth = static_cast<std::size_t> (base_delta / sigma_r_) + 1 + 2 * padding_z;
110 
111  Array3D data (small_width, small_height, small_depth);
112 #if OPENMP_LEGACY_CONST_DATA_SHARING_RULE
113 #pragma omp parallel for \
114  default(none) \
115  shared(base_min, data, output) \
116  num_threads(threads_)
117 #else
118 #pragma omp parallel for \
119  default(none) \
120  shared(base_min, data, output, small_height, small_width) \
121  num_threads(threads_)
122 #endif
123  for (long int i = 0; i < static_cast<long int> (small_width * small_height); ++i)
124  {
125  std::size_t small_x = static_cast<std::size_t> (i % small_width);
126  std::size_t small_y = static_cast<std::size_t> (i / small_width);
127  std::size_t start_x = static_cast<std::size_t>(
128  std::max ((static_cast<float> (small_x) - static_cast<float> (padding_xy) - 0.5f) * sigma_s_ + 1, 0.f));
129  std::size_t end_x = static_cast<std::size_t>(
130  std::max ((static_cast<float> (small_x) - static_cast<float> (padding_xy) + 0.5f) * sigma_s_ + 1, 0.f));
131  std::size_t start_y = static_cast<std::size_t>(
132  std::max ((static_cast<float> (small_y) - static_cast<float> (padding_xy) - 0.5f) * sigma_s_ + 1, 0.f));
133  std::size_t end_y = static_cast<std::size_t>(
134  std::max ((static_cast<float> (small_y) - static_cast<float> (padding_xy) + 0.5f) * sigma_s_ + 1, 0.f));
135  for (std::size_t x = start_x; x < end_x && x < input_->width; ++x)
136  {
137  for (std::size_t y = start_y; y < end_y && y < input_->height; ++y)
138  {
139  const float z = output (x,y).z - base_min;
140  const std::size_t small_z = static_cast<std::size_t> (static_cast<float> (z) / sigma_r_ + 0.5f) + padding_z;
141  Eigen::Vector2f& d = data (small_x, small_y, small_z);
142  d[0] += output (x,y).z;
143  d[1] += 1.0f;
144  }
145  }
146  }
147 
148  std::vector<long int> offset (3);
149  offset[0] = &(data (1,0,0)) - &(data (0,0,0));
150  offset[1] = &(data (0,1,0)) - &(data (0,0,0));
151  offset[2] = &(data (0,0,1)) - &(data (0,0,0));
152 
153  Array3D buffer (small_width, small_height, small_depth);
154 
155  for (std::size_t dim = 0; dim < 3; ++dim)
156  {
157  for (std::size_t n_iter = 0; n_iter < 2; ++n_iter)
158  {
159  Array3D* current_buffer = (n_iter % 2 == 1 ? &buffer : &data);
160  Array3D* current_data =(n_iter % 2 == 1 ? &data : &buffer);
161 #if OPENMP_LEGACY_CONST_DATA_SHARING_RULE
162 #pragma omp parallel for \
163  default(none) \
164  shared(current_buffer, current_data, dim, offset) \
165  num_threads(threads_)
166 #else
167 #pragma omp parallel for \
168  default(none) \
169  shared(current_buffer, current_data, dim, offset, small_depth, small_height, small_width) \
170  num_threads(threads_)
171 #endif
172  for(long int i = 0; i < static_cast<long int> ((small_width - 2)*(small_height - 2)); ++i)
173  {
174  std::size_t x = static_cast<std::size_t> (i % (small_width - 2) + 1);
175  std::size_t y = static_cast<std::size_t> (i / (small_width - 2) + 1);
176  const long int off = offset[dim];
177  Eigen::Vector2f* d_ptr = &(current_data->operator() (x,y,1));
178  Eigen::Vector2f* b_ptr = &(current_buffer->operator() (x,y,1));
179 
180  for(std::size_t z = 1; z < small_depth - 1; ++z, ++d_ptr, ++b_ptr)
181  *d_ptr = (*(b_ptr - off) + *(b_ptr + off) + 2.0 * (*b_ptr)) / 4.0;
182  }
183  }
184  }
185  // Note: this works because there are an even number of iterations.
186  // If there were an odd number, we would need to end with a:
187  // std::swap (data, buffer);
188 
189  if (early_division_)
190  {
191  for (std::vector<Eigen::Vector2f, Eigen::aligned_allocator<Eigen::Vector2f> >::iterator d = data.begin (); d != data.end (); ++d)
192  *d /= ((*d)[0] != 0) ? (*d)[1] : 1;
193 
194 #pragma omp parallel for \
195  default(none) \
196  shared(base_min, data, output) \
197  num_threads(threads_)
198  for (long int i = 0; i < static_cast<long int> (input_->size ()); ++i)
199  {
200  std::size_t x = static_cast<std::size_t> (i % input_->width);
201  std::size_t y = static_cast<std::size_t> (i / input_->width);
202  const float z = output (x,y).z - base_min;
203  const Eigen::Vector2f D = data.trilinear_interpolation (static_cast<float> (x) / sigma_s_ + padding_xy,
204  static_cast<float> (y) / sigma_s_ + padding_xy,
205  z / sigma_r_ + padding_z);
206  output(x,y).z = D[0];
207  }
208  }
209  else
210  {
211 #pragma omp parallel for \
212  default(none) \
213  shared(base_min, data, output) \
214  num_threads(threads_)
215  for (long i = 0; i < static_cast<long int> (input_->size ()); ++i)
216  {
217  std::size_t x = static_cast<std::size_t> (i % input_->width);
218  std::size_t y = static_cast<std::size_t> (i / input_->width);
219  const float z = output (x,y).z - base_min;
220  const Eigen::Vector2f D = data.trilinear_interpolation (static_cast<float> (x) / sigma_s_ + padding_xy,
221  static_cast<float> (y) / sigma_s_ + padding_xy,
222  z / sigma_r_ + padding_z);
223  output (x,y).z = D[0] / D[1];
224  }
225  }
226 }
227 
228 
229 
230 #endif /* PCL_FILTERS_IMPL_FAST_BILATERAL_OMP_HPP_ */
231 
pcl::PointCloud::height
std::uint32_t height
The point cloud height (if organized as an image-structure).
Definition: point_cloud.h:415
pcl::PointCloud
PointCloud represents the base class in PCL for storing collections of 3D points.
Definition: projection_matrix.h:52
pcl::FastBilateralFilterOMP::setNumberOfThreads
void setNumberOfThreads(unsigned int nr_threads=0)
Initialize the scheduler and set the number of threads to use.
Definition: fast_bilateral_omp.hpp:49
pcl::FastBilateralFilterOMP::Array3D
typename FastBilateralFilter< PointT >::Array3D Array3D
Definition: fast_bilateral_omp.h:64
pcl::copyPointCloud
void copyPointCloud(const pcl::PointCloud< PointInT > &cloud_in, pcl::PointCloud< PointOutT > &cloud_out)
Copy all the fields from a given point cloud into a new point cloud.
Definition: io.hpp:121
pcl::PointCloud::at
const PointT & at(int column, int row) const
Obtain the point given by the (column, row) coordinates.
Definition: point_cloud.h:283
pcl::PointCloud::width
std::uint32_t width
The point cloud width (if organized as an image-structure).
Definition: point_cloud.h:413
pcl::FastBilateralFilterOMP::applyFilter
void applyFilter(PointCloud &output) override
Filter the input data and store the results into output.
Definition: fast_bilateral_omp.hpp:63
pcl::PointCloud::size
std::size_t size() const
Definition: point_cloud.h:448