NearestNeighborsFLANN.h
1 /*********************************************************************
2 * Software License Agreement (BSD License)
3 *
4 * Copyright (c) 2011, Rice University
5 * All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 *
11 * * Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * * Redistributions in binary form must reproduce the above
14 * copyright notice, this list of conditions and the following
15 * disclaimer in the documentation and/or other materials provided
16 * with the distribution.
17 * * Neither the name of the Rice University nor the names of its
18 * contributors may be used to endorse or promote products derived
19 * from this software without specific prior written permission.
20 *
21 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
22 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
23 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
24 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
25 * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
26 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
27 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
28 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
29 * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
30 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
31 * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
32 * POSSIBILITY OF SUCH DAMAGE.
33 *********************************************************************/
34 
35 /* Author: Mark Moll */
36 
37 #ifndef OMPL_DATASTRUCTURES_NEAREST_NEIGHBORS_FLANN_
38 #define OMPL_DATASTRUCTURES_NEAREST_NEIGHBORS_FLANN_
39 
40 #include "ompl/config.h"
41 #if OMPL_HAVE_FLANN == 0
42 #error FLANN is not available. Please use a different NearestNeighbors data structure.
43 #else
44 
45 #include "ompl/datastructures/NearestNeighbors.h"
46 #include "ompl/util/Exception.h"
47 
48 #include <flann/flann.hpp>
49 #include <utility>
50 
51 namespace ompl
52 {
56  template <typename _T>
57  class FLANNDistance
58  {
59  public:
60  using ElementType = _T;
61  using ResultType = double;
62 
63  FLANNDistance(const typename NearestNeighbors<_T>::DistanceFunction &distFun) : distFun_(distFun)
64  {
65  }
66 
67  template <typename Iterator1, typename Iterator2>
68  ResultType operator()(Iterator1 a, Iterator2 b, size_t /*size*/, ResultType /*worst_dist*/ = -1) const
69  {
70  return distFun_(*a, *b);
71  }
72 
73  protected:
74  const typename NearestNeighbors<_T>::DistanceFunction &distFun_;
75  };
76 
86  template <typename _T, typename _Dist = FLANNDistance<_T>>
87  class NearestNeighborsFLANN : public NearestNeighbors<_T>
88  {
89  public:
90  NearestNeighborsFLANN(std::shared_ptr<flann::IndexParams> params)
91  : index_(nullptr), params_(std::move(params)), searchParams_(32, 0., true), dimension_(1)
92  {
93  }
94 
95  ~NearestNeighborsFLANN() override
96  {
97  if (index_)
98  delete index_;
99  }
100 
101  void clear() override
102  {
103  if (index_)
104  {
105  delete index_;
106  index_ = nullptr;
107  }
108  data_.clear();
109  }
110 
111  bool reportsSortedResults() const override
112  {
113  return searchParams_.sorted;
114  }
115 
116  void setDistanceFunction(const typename NearestNeighbors<_T>::DistanceFunction &distFun) override
117  {
120  }
121 
122  void add(const _T &data) override
123  {
124  bool rebuild = index_ && (data_.size() + 1 > data_.capacity());
125 
126  if (rebuild)
127  rebuildIndex(2 * data_.capacity());
128 
129  data_.push_back(data);
130  const flann::Matrix<_T> mat(&data_.back(), 1, dimension_);
131 
132  if (index_)
133  index_->addPoints(mat, std::numeric_limits<float>::max() / size());
134  else
135  createIndex(mat);
136  }
137  void add(const std::vector<_T> &data) override
138  {
139  if (data.empty()) return;
140  unsigned int oldSize = data_.size();
141  unsigned int newSize = oldSize + data.size();
142  bool rebuild = index_ && (newSize > data_.capacity());
143 
144  if (rebuild)
145  rebuildIndex(std::max(2 * oldSize, newSize));
146 
147  if (index_)
148  {
149  std::copy(data.begin(), data.end(), data_.begin() + oldSize);
150  const flann::Matrix<_T> mat(&data_[oldSize], data.size(), dimension_);
151  index_->addPoints(mat, std::numeric_limits<float>::max() / size());
152  }
153  else
154  {
155  data_ = data;
156  const flann::Matrix<_T> mat(&data_[0], data_.size(), dimension_);
157  createIndex(mat);
158  }
159  }
160  bool remove(const _T &data) override
161  {
162  if (!index_)
163  return false;
164  auto &elt = const_cast<_T &>(data);
165  const flann::Matrix<_T> query(&elt, 1, dimension_);
166  std::vector<std::vector<size_t>> indices(1);
167  std::vector<std::vector<double>> dists(1);
168  index_->knnSearch(query, indices, dists, 1, searchParams_);
169  if (*index_->getPoint(indices[0][0]) == data)
170  {
171  index_->removePoint(indices[0][0]);
172  rebuildIndex();
173  return true;
174  }
175  return false;
176  }
177  _T nearest(const _T &data) const override
178  {
179  if (size())
180  {
181  auto &elt = const_cast<_T &>(data);
182  const flann::Matrix<_T> query(&elt, 1, dimension_);
183  std::vector<std::vector<size_t>> indices(1);
184  std::vector<std::vector<double>> dists(1);
185  index_->knnSearch(query, indices, dists, 1, searchParams_);
186  return *index_->getPoint(indices[0][0]);
187  }
188  throw Exception("No elements found in nearest neighbors data structure");
189  }
192  void nearestK(const _T &data, std::size_t k, std::vector<_T> &nbh) const override
193  {
194  auto &elt = const_cast<_T &>(data);
195  const flann::Matrix<_T> query(&elt, 1, dimension_);
196  std::vector<std::vector<size_t>> indices;
197  std::vector<std::vector<double>> dists;
198  k = index_ ? index_->knnSearch(query, indices, dists, k, searchParams_) : 0;
199  nbh.resize(k);
200  for (std::size_t i = 0; i < k; ++i)
201  nbh[i] = *index_->getPoint(indices[0][i]);
202  }
205  void nearestR(const _T &data, double radius, std::vector<_T> &nbh) const override
206  {
207  auto &elt = const_cast<_T &>(data);
208  flann::Matrix<_T> query(&elt, 1, dimension_);
209  std::vector<std::vector<size_t>> indices;
210  std::vector<std::vector<double>> dists;
211  int k = index_ ? index_->radiusSearch(query, indices, dists, radius, searchParams_) : 0;
212  nbh.resize(k);
213  for (int i = 0; i < k; ++i)
214  nbh[i] = *index_->getPoint(indices[0][i]);
215  }
216 
217  std::size_t size() const override
218  {
219  return index_ ? index_->size() : 0;
220  }
221 
222  void list(std::vector<_T> &data) const override
223  {
224  std::size_t sz = size();
225  if (sz == 0)
226  {
227  data.resize(0);
228  return;
229  }
230  const _T &dummy = *index_->getPoint(0);
231  int checks = searchParams_.checks;
232  searchParams_.checks = size();
233  nearestK(dummy, sz, data);
234  searchParams_.checks = checks;
235  }
236 
241  virtual void setIndexParams(const std::shared_ptr<flann::IndexParams> &params)
242  {
243  params_ = params;
244  rebuildIndex();
245  }
246 
248  virtual const std::shared_ptr<flann::IndexParams> &getIndexParams() const
249  {
250  return params_;
251  }
252 
255  virtual void setSearchParams(const flann::SearchParams &searchParams)
256  {
257  searchParams_ = searchParams;
258  }
259 
262  flann::SearchParams &getSearchParams()
263  {
264  return searchParams_;
265  }
266 
269  const flann::SearchParams &getSearchParams() const
270  {
271  return searchParams_;
272  }
273 
274  unsigned int getContainerSize() const
275  {
276  return dimension_;
277  }
278 
279  protected:
282  void createIndex(const flann::Matrix<_T> &mat)
283  {
284  index_ = new flann::Index<_Dist>(mat, *params_, _Dist(NearestNeighbors<_T>::distFun_));
285  index_->buildIndex();
286  }
287 
290  void rebuildIndex(unsigned int capacity = 0)
291  {
292  if (index_)
293  {
294  std::vector<_T> data;
295  list(data);
296  clear();
297  if (capacity != 0u)
298  data_.reserve(capacity);
299  add(data);
300  }
301  }
302 
305  std::vector<_T> data_;
306 
308  flann::Index<_Dist> *index_;
309 
312  std::shared_ptr<flann::IndexParams> params_;
313 
315  mutable flann::SearchParams searchParams_;
316 
320  unsigned int dimension_;
321  };
322 
323  template <>
324  inline void NearestNeighborsFLANN<double, flann::L2<double>>::createIndex(
325  const flann::Matrix<double> &mat)
326  {
327  index_ = new flann::Index<flann::L2<double>>(mat, *params_);
328  index_->buildIndex();
329  }
330 
331  template <typename _T, typename _Dist = FLANNDistance<_T>>
332  class NearestNeighborsFLANNLinear : public NearestNeighborsFLANN<_T, _Dist>
333  {
334  public:
336  : NearestNeighborsFLANN<_T, _Dist>(std::shared_ptr<flann::LinearIndexParams>(new flann::LinearIndexParams()))
337  {
338  }
339  };
340 
341  template <typename _T, typename _Dist = FLANNDistance<_T>>
343  {
344  public:
346  : NearestNeighborsFLANN<_T, _Dist>(std::shared_ptr<flann::HierarchicalClusteringIndexParams>(
347  new flann::HierarchicalClusteringIndexParams()))
348  {
349  }
350  };
351 }
352 #endif
353 
354 #endif
void nearestR(const _T &data, double radius, std::vector< _T > &nbh) const override
Return the nearest neighbors within distance radius in sorted order if searchParams_....
virtual void setSearchParams(const flann::SearchParams &searchParams)
Set the FLANN parameters to be used during nearest neighbor searches.
void add(const _T &data) override
Add an element to the datastructure.
flann::SearchParams searchParams_
The parameters used to seach for nearest neighbors.
bool remove(const _T &data) override
Remove an element from the datastructure.
std::function< double(const _T &, const _T &)> DistanceFunction
The definition of a distance function.
void rebuildIndex(unsigned int capacity=0)
Rebuild the nearest neighbor data structure (necessary when changing the distance function or index p...
flann::Index< _Dist > * index_
The FLANN index (the actual index type depends on params_).
void createIndex(const flann::Matrix< _T > &mat)
Internal function to construct nearest neighbor data structure with initial elements stored in mat.
_T nearest(const _T &data) const override
Get the nearest neighbor of a point.
Abstract representation of a container that can perform nearest neighbors queries.
unsigned int dimension_
If each element has an array-like structure that is exposed to FLANN, then the dimension_ needs to be...
bool reportsSortedResults() const override
Return true if the solutions reported by this data structure are sorted, when calling nearestK / near...
void list(std::vector< _T > &data) const override
Get all the elements in the datastructure.
void clear() override
Clear the datastructure.
Wrapper class for nearest neighbor data structures in the FLANN library.
flann::SearchParams & getSearchParams()
Get the FLANN parameters used during nearest neighbor searches.
virtual void setIndexParams(const std::shared_ptr< flann::IndexParams > &params)
Set the FLANN index parameters.
std::size_t size() const override
Get the number of elements in the datastructure.
std::vector< _T > data_
vector of data stored in FLANN's index. FLANN only indexes references, so we need store the original ...
std::shared_ptr< flann::IndexParams > params_
The FLANN index parameters. This contains both the type of index and the parameters for that type.
virtual const std::shared_ptr< flann::IndexParams > & getIndexParams() const
Get the FLANN parameters used to build the current index.
void nearestK(const _T &data, std::size_t k, std::vector< _T > &nbh) const override
Return the k nearest neighbors in sorted order if searchParams_.sorted==true (the default)
Main namespace. Contains everything in this library.
virtual void setDistanceFunction(const DistanceFunction &distFun)
Set the distance function to use.