mrpt::math::KDTreeCapable< Derived, num_t, metric_t > Class Template Reference

Detailed Description

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>
class mrpt::math::KDTreeCapable< Derived, num_t, metric_t >

A generic adaptor class for providing Nearest Neighbor (NN) lookup via the nanoflann library.

This makes use of the CRTP design pattern.

Derived classes must be aware of the need to call "kdtree_mark_as_outdated()" when the data points change to mark the cached KD-tree (an "index") as invalid, and also implement the following interface (note that these are not virtual functions due to the usage of CRTP):

  // Must return the number of data points
  inline size_t kdtree_get_point_count() const { ... }

  // Returns the distance between the vector "p1[0:size-1]" and the data
point with index "idx_p2" stored in the class:
  inline float kdtree_distance(const float *p1, const size_t idx_p2,size_t
size) const { ... }

  // Returns the dim'th component of the idx'th point in the class:
  inline num_t kdtree_get_pt(const size_t idx, int dim) const { ... }

  // Optional bounding-box computation: return false to default to a standard
bbox computation loop.
  //   Return true if the BBOX was already computed by the class and returned
in "bb" so it can be avoided to redo it again.
  //   Look at bb.size() to find out the expected dimensionality (e.g. 2 or 3
for point clouds)
  template <class BBOX>
  bool kdtree_get_bbox(BBOX &bb) const
  {
     bb[0].low = ...; bb[0].high = ...;  // "x" limits
     return true;
  }

The KD-tree index will be built on demand only upon call of any of the query methods provided by this class.

Notice that there is only ONE internal cached KD-tree, so if a method to query a 2D point is called, then another method for 3D points, then again the 2D method, three KD-trees will be built. So, try to group all the calls for a given dimensionality together or build different class instances for queries of each dimensionality, etc.

See also

See some of the derived classes for example implementations. See also the documentation of nanoflann

Definition at line 83 of file KDTreeCapable.h.

#include <mrpt/math/KDTreeCapable.h>

Inheritance diagram for mrpt::math::KDTreeCapable< Derived, num_t, metric_t >:

Classes
struct	TKDTreeDataHolder
	Internal structure with the KD-tree representation (mainly used to avoid copying pointers with the = operator) More...
struct	TKDTreeSearchParams

Public Types
using	self_t = KDTreeCapable< Derived, num_t, metric_t >

Public Member Functions
	KDTreeCapable ()=default
	Constructor. More...
	KDTreeCapable (const KDTreeCapable &)
KDTreeCapable &	operator= (const KDTreeCapable &)
const Derived &	derived () const
	CRTP helper method. More...
Derived &	derived ()
	CRTP helper method. More...
Public utility methods to query the KD-tree
size_t	kdTreeClosestPoint2D (float x0, float y0, float &out_x, float &out_y, float &out_dist_sqr) const
	KD Tree-based search for the closest point (only ONE) to some given 2D coordinates. More...
size_t	kdTreeClosestPoint2D (float x0, float y0, float &out_dist_sqr) const
size_t	kdTreeClosestPoint2D (const TPoint2D &p0, TPoint2D &pOut, float &outDistSqr) const
float	kdTreeClosestPoint2DsqrError (float x0, float y0) const
	Like kdTreeClosestPoint2D, but just return the square error from some point to its closest neighbor. More...
float	kdTreeClosestPoint2DsqrError (const TPoint2D &p0) const
void	kdTreeTwoClosestPoint2D (float x0, float y0, float &out_x1, float &out_y1, float &out_x2, float &out_y2, float &out_dist_sqr1, float &out_dist_sqr2) const
	KD Tree-based search for the TWO closest point to some given 2D coordinates. More...
void	kdTreeTwoClosestPoint2D (const TPoint2D &p0, TPoint2D &pOut1, TPoint2D &pOut2, float &outDistSqr1, float &outDistSqr2) const
std::vector< size_t >	kdTreeNClosestPoint2D (float x0, float y0, size_t knn, std::vector< float > &out_x, std::vector< float > &out_y, std::vector< float > &out_dist_sqr) const
	KD Tree-based search for the N closest point to some given 2D coordinates. More...
std::vector< size_t >	kdTreeNClosestPoint2D (const TPoint2D &p0, size_t N, std::vector< TPoint2D > &pOut, std::vector< float > &outDistSqr) const
void	kdTreeNClosestPoint2DIdx (float x0, float y0, size_t knn, std::vector< size_t > &out_idx, std::vector< float > &out_dist_sqr) const
	KD Tree-based search for the N closest point to some given 2D coordinates and returns their indexes. More...
void	kdTreeNClosestPoint2DIdx (const TPoint2D &p0, size_t N, std::vector< size_t > &outIdx, std::vector< float > &outDistSqr) const
size_t	kdTreeClosestPoint3D (float x0, float y0, float z0, float &out_x, float &out_y, float &out_z, float &out_dist_sqr) const
	KD Tree-based search for the closest point (only ONE) to some given 3D coordinates. More...
size_t	kdTreeClosestPoint3D (float x0, float y0, float z0, float &out_dist_sqr) const
size_t	kdTreeClosestPoint3D (const TPoint3D &p0, TPoint3D &pOut, float &outDistSqr) const
void	kdTreeNClosestPoint3D (float x0, float y0, float z0, size_t knn, std::vector< float > &out_x, std::vector< float > &out_y, std::vector< float > &out_z, std::vector< float > &out_dist_sqr) const
	KD Tree-based search for the N closest points to some given 3D coordinates. More...
void	kdTreeNClosestPoint3DWithIdx (float x0, float y0, float z0, size_t knn, std::vector< float > &out_x, std::vector< float > &out_y, std::vector< float > &out_z, std::vector< size_t > &out_idx, std::vector< float > &out_dist_sqr) const
	KD Tree-based search for the N closest points to some given 3D coordinates. More...
void	kdTreeNClosestPoint3D (const TPoint3D &p0, size_t N, std::vector< TPoint3D > &pOut, std::vector< float > &outDistSqr) const
size_t	kdTreeRadiusSearch3D (const num_t x0, const num_t y0, const num_t z0, const num_t maxRadiusSqr, std::vector< std::pair< size_t, num_t >> &out_indices_dist) const
	KD Tree-based search for all the points within a given radius of some 3D point. More...
size_t	kdTreeRadiusSearch2D (const num_t x0, const num_t y0, const num_t maxRadiusSqr, std::vector< std::pair< size_t, num_t >> &out_indices_dist) const
	KD Tree-based search for all the points within a given radius of some 2D point. More...
void	kdTreeNClosestPoint3DIdx (float x0, float y0, float z0, size_t knn, std::vector< size_t > &out_idx, std::vector< float > &out_dist_sqr) const
	KD Tree-based search for the N closest point to some given 3D coordinates and returns their indexes. More...
void	kdTreeNClosestPoint3DIdx (const TPoint3D &p0, size_t N, std::vector< size_t > &outIdx, std::vector< float > &outDistSqr) const
void	kdTreeEnsureIndexBuilt3D ()
void	kdTreeEnsureIndexBuilt2D ()

Public Attributes
TKDTreeSearchParams	kdtree_search_params
	Parameters to tune the ANN searches. More...

Protected Member Functions
void	kdtree_mark_as_outdated () const
	To be called by child classes when KD tree data changes. More...

Private Member Functions
void	rebuild_kdTree_2D () const
	Rebuild, if needed the KD-tree for 2D (nDims=2), 3D (nDims=3), ... More...
void	rebuild_kdTree_3D () const
	Rebuild, if needed the KD-tree for 2D (nDims=2), 3D (nDims=3), ... More...

Private Attributes
std::mutex	m_kdtree_mtx
TKDTreeDataHolder< 2 >	m_kdtree2d_data
TKDTreeDataHolder< 3 >	m_kdtree3d_data
std::atomic_bool	m_kdtree_is_uptodate {false}
	whether the KD tree needs to be rebuilt or not. More...

Member Typedef Documentation

self_t

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

using mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::self_t = KDTreeCapable<Derived, num_t, metric_t>

Definition at line 87 of file KDTreeCapable.h.

Constructor & Destructor Documentation

KDTreeCapable() [1/2]

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::KDTreeCapable ( )

inlinedefault

Constructor.

KDTreeCapable() [2/2]

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::KDTreeCapable ( const KDTreeCapable< Derived, num_t, metric_t > &  )

inline

Definition at line 92 of file KDTreeCapable.h.

Member Function Documentation

derived() [1/2]

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

const Derived& mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::derived ( ) const

inline

CRTP helper method.

Definition at line 100 of file KDTreeCapable.h.

Referenced by mrpt::math::KDTreeCapable< CFeatureList >::kdTreeClosestPoint2D(), mrpt::math::KDTreeCapable< CFeatureList >::kdTreeClosestPoint3D(), mrpt::math::KDTreeCapable< CFeatureList >::kdTreeNClosestPoint2D(), mrpt::math::KDTreeCapable< CFeatureList >::kdTreeNClosestPoint3D(), mrpt::math::KDTreeCapable< CFeatureList >::kdTreeNClosestPoint3DWithIdx(), mrpt::math::KDTreeCapable< CFeatureList >::kdTreeTwoClosestPoint2D(), mrpt::math::KDTreeCapable< CFeatureList >::rebuild_kdTree_2D(), and mrpt::math::KDTreeCapable< CFeatureList >::rebuild_kdTree_3D().

Here is the caller graph for this function:

derived() [2/2]

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

Derived& mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::derived ( )

inline

CRTP helper method.

Definition at line 105 of file KDTreeCapable.h.

kdtree_mark_as_outdated()

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

void mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::kdtree_mark_as_outdated ( ) const

inlineprotected

To be called by child classes when KD tree data changes.

Definition at line 721 of file KDTreeCapable.h.

Referenced by mrpt::math::KDTreeCapable< CFeatureList >::operator=().

Here is the caller graph for this function:

kdTreeClosestPoint2D() [1/3]

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

size_t mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::kdTreeClosestPoint2D ( float  x0, float  y0, float &  out_x, float &  out_y, float &  out_dist_sqr  ) const

inline

KD Tree-based search for the closest point (only ONE) to some given 2D coordinates.

This method automatically build the "m_kdtree_data" structure when:

• It is called for the first time
• The map has changed
• The KD-tree was build for 3D.

Parameters

x0	The X coordinate of the query.
y0	The Y coordinate of the query.
out_x	The X coordinate of the found closest correspondence.
out_y	The Y coordinate of the found closest correspondence.
out_dist_sqr	The square distance between the query and the returned point.

Returns

The index of the closest point in the map array.

See also

kdTreeClosestPoint3D, kdTreeTwoClosestPoint2D

Definition at line 136 of file KDTreeCapable.h.

Referenced by mrpt::math::KDTreeCapable< CFeatureList >::kdTreeClosestPoint2D(), and mrpt::math::KDTreeCapable< CFeatureList >::kdTreeClosestPoint2DsqrError().

Here is the caller graph for this function:

kdTreeClosestPoint2D() [2/3]

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

size_t mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::kdTreeClosestPoint2D ( float  x0, float  y0, float &  out_dist_sqr  ) const

inline

This is an overloaded member function, provided for convenience. It differs from the above function only in what argument(s) it accepts.

Definition at line 163 of file KDTreeCapable.h.

kdTreeClosestPoint2D() [3/3]

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

size_t mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::kdTreeClosestPoint2D ( const TPoint2D &  p0, TPoint2D &  pOut, float &  outDistSqr  ) const

inline

This is an overloaded member function, provided for convenience. It differs from the above function only in what argument(s) it accepts.

Definition at line 185 of file KDTreeCapable.h.

kdTreeClosestPoint2DsqrError() [1/2]

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

float mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::kdTreeClosestPoint2DsqrError ( float  x0, float  y0  ) const

inline

Like kdTreeClosestPoint2D, but just return the square error from some point to its closest neighbor.

Definition at line 200 of file KDTreeCapable.h.

Referenced by mrpt::math::KDTreeCapable< CFeatureList >::kdTreeClosestPoint2DsqrError().

Here is the caller graph for this function:

kdTreeClosestPoint2DsqrError() [2/2]

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

float mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::kdTreeClosestPoint2DsqrError ( const TPoint2D &  p0 ) const

inline

Definition at line 207 of file KDTreeCapable.h.

kdTreeClosestPoint3D() [1/3]

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

size_t mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::kdTreeClosestPoint3D ( float  x0, float  y0, float  z0, float &  out_x, float &  out_y, float &  out_z, float &  out_dist_sqr  ) const

inline

KD Tree-based search for the closest point (only ONE) to some given 3D coordinates.

This method automatically build the "m_kdtree_data" structure when:

• It is called for the first time
• The map has changed
• The KD-tree was build for 2D.

Parameters

x0	The X coordinate of the query.
y0	The Y coordinate of the query.
z0	The Z coordinate of the query.
out_x	The X coordinate of the found closest correspondence.
out_y	The Y coordinate of the found closest correspondence.
out_z	The Z coordinate of the found closest correspondence.
out_dist_sqr	The square distance between the query and the returned point.

Returns

The index of the closest point in the map array.

See also

kdTreeClosestPoint2D

Definition at line 409 of file KDTreeCapable.h.

Referenced by mrpt::math::KDTreeCapable< CFeatureList >::kdTreeClosestPoint3D().

Here is the caller graph for this function:

kdTreeClosestPoint3D() [2/3]

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

size_t mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::kdTreeClosestPoint3D ( float  x0, float  y0, float  z0, float &  out_dist_sqr  ) const

inline

This is an overloaded member function, provided for convenience. It differs from the above function only in what argument(s) it accepts.

Definition at line 437 of file KDTreeCapable.h.

kdTreeClosestPoint3D() [3/3]

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

size_t mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::kdTreeClosestPoint3D ( const TPoint3D &  p0, TPoint3D &  pOut, float &  outDistSqr  ) const

inline

This is an overloaded member function, provided for convenience. It differs from the above function only in what argument(s) it accepts.

Definition at line 459 of file KDTreeCapable.h.

kdTreeEnsureIndexBuilt2D()

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

void mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::kdTreeEnsureIndexBuilt2D ( )

inline

Definition at line 715 of file KDTreeCapable.h.

kdTreeEnsureIndexBuilt3D()

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

void mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::kdTreeEnsureIndexBuilt3D ( )

inline

Definition at line 714 of file KDTreeCapable.h.

kdTreeNClosestPoint2D() [1/2]

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

std::vector<size_t> mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::kdTreeNClosestPoint2D ( float  x0, float  y0, size_t  knn, std::vector< float > &  out_x, std::vector< float > &  out_y, std::vector< float > &  out_dist_sqr  ) const

inline

KD Tree-based search for the N closest point to some given 2D coordinates.

This method automatically build the "m_kdtree_data" structure when:

• It is called for the first time
• The map has changed
• The KD-tree was build for 3D.

Parameters

x0	The X coordinate of the query.
y0	The Y coordinate of the query.
N	The number of closest points to search.
out_x	The vector containing the X coordinates of the correspondences.
out_y	The vector containing the Y coordinates of the correspondences.
out_dist_sqr	The vector containing the square distance between the query and the returned points.

Returns

The list of indices

See also

kdTreeClosestPoint2D

kdTreeTwoClosestPoint2D

Definition at line 298 of file KDTreeCapable.h.

Referenced by mrpt::math::KDTreeCapable< CFeatureList >::kdTreeNClosestPoint2D().

Here is the caller graph for this function:

kdTreeNClosestPoint2D() [2/2]

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

std::vector<size_t> mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::kdTreeNClosestPoint2D ( const TPoint2D &  p0, size_t  N, std::vector< TPoint2D > &  pOut, std::vector< float > &  outDistSqr  ) const

inline

Definition at line 328 of file KDTreeCapable.h.

kdTreeNClosestPoint2DIdx() [1/2]

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

void mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::kdTreeNClosestPoint2DIdx ( float  x0, float  y0, size_t  knn, std::vector< size_t > &  out_idx, std::vector< float > &  out_dist_sqr  ) const

inline

KD Tree-based search for the N closest point to some given 2D coordinates and returns their indexes.

This method automatically build the "m_kdtree_data" structure when:

• It is called for the first time
• The map has changed
• The KD-tree was build for 3D.

Parameters

x0	The X coordinate of the query.
y0	The Y coordinate of the query.
N	The number of closest points to search.
out_idx	The indexes of the found closest correspondence.
out_dist_sqr	The square distance between the query and the returned point.

See also

kdTreeClosestPoint2D

Definition at line 361 of file KDTreeCapable.h.

Referenced by mrpt::slam::CGridMapAligner::AlignPDF_robustMatch(), and mrpt::math::KDTreeCapable< CFeatureList >::kdTreeNClosestPoint2DIdx().

Here is the caller graph for this function:

kdTreeNClosestPoint2DIdx() [2/2]

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

void mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::kdTreeNClosestPoint2DIdx ( const TPoint2D &  p0, size_t  N, std::vector< size_t > &  outIdx, std::vector< float > &  outDistSqr  ) const

inline

Definition at line 381 of file KDTreeCapable.h.

kdTreeNClosestPoint3D() [1/2]

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

void mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::kdTreeNClosestPoint3D ( float  x0, float  y0, float  z0, size_t  knn, std::vector< float > &  out_x, std::vector< float > &  out_y, std::vector< float > &  out_z, std::vector< float > &  out_dist_sqr  ) const

inline

KD Tree-based search for the N closest points to some given 3D coordinates.

This method automatically build the "m_kdtree_data" structure when:

• It is called for the first time
• The map has changed
• The KD-tree was build for 2D.

Parameters

x0	The X coordinate of the query.
y0	The Y coordinate of the query.
z0	The Z coordinate of the query.
N	The number of closest points to search.
out_x	The vector containing the X coordinates of the correspondences.
out_y	The vector containing the Y coordinates of the correspondences.
out_z	The vector containing the Z coordinates of the correspondences.
out_dist_sqr	The vector containing the square distance between the query and the returned points.

See also

kdTreeNClosestPoint2D

Definition at line 494 of file KDTreeCapable.h.

Referenced by mrpt::maps::CPointCloudFilterByDistance::filter(), and mrpt::math::KDTreeCapable< CFeatureList >::kdTreeNClosestPoint3D().

Here is the caller graph for this function:

kdTreeNClosestPoint3D() [2/2]

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

void mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::kdTreeNClosestPoint3D ( const TPoint3D &  p0, size_t  N, std::vector< TPoint3D > &  pOut, std::vector< float > &  outDistSqr  ) const

inline

Definition at line 581 of file KDTreeCapable.h.

kdTreeNClosestPoint3DIdx() [1/2]

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

void mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::kdTreeNClosestPoint3DIdx ( float  x0, float  y0, float  z0, size_t  knn, std::vector< size_t > &  out_idx, std::vector< float > &  out_dist_sqr  ) const

inline

KD Tree-based search for the N closest point to some given 3D coordinates and returns their indexes.

This method automatically build the "m_kdtree_data" structure when:

• It is called for the first time
• The map has changed
• The KD-tree was build for 2D.

Parameters

x0	The X coordinate of the query.
y0	The Y coordinate of the query.
z0	The Z coordinate of the query.
N	The number of closest points to search.
out_idx	The indexes of the found closest correspondence.
out_dist_sqr	The square distance between the query and the returned point.

See also

kdTreeClosestPoint2D, kdTreeRadiusSearch3D

Definition at line 685 of file KDTreeCapable.h.

Referenced by mrpt::math::KDTreeCapable< CFeatureList >::kdTreeNClosestPoint3DIdx().

Here is the caller graph for this function:

kdTreeNClosestPoint3DIdx() [2/2]

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

void mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::kdTreeNClosestPoint3DIdx ( const TPoint3D &  p0, size_t  N, std::vector< size_t > &  outIdx, std::vector< float > &  outDistSqr  ) const

inline

Definition at line 705 of file KDTreeCapable.h.

kdTreeNClosestPoint3DWithIdx()

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

void mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::kdTreeNClosestPoint3DWithIdx ( float  x0, float  y0, float  z0, size_t  knn, std::vector< float > &  out_x, std::vector< float > &  out_y, std::vector< float > &  out_z, std::vector< size_t > &  out_idx, std::vector< float > &  out_dist_sqr  ) const

inline

KD Tree-based search for the N closest points to some given 3D coordinates.

This method automatically build the "m_kdtree_data" structure when:

• It is called for the first time
• The map has changed
• The KD-tree was build for 2D.

Parameters

x0	The X coordinate of the query.
y0	The Y coordinate of the query.
z0	The Z coordinate of the query.
N	The number of closest points to search.
out_x	The vector containing the X coordinates of the correspondences.
out_y	The vector containing the Y coordinates of the correspondences.
out_z	The vector containing the Z coordinates of the correspondences.
out_idx	The vector containing the indexes of the correspondences.
out_dist_sqr	The vector containing the square distance between the query and the returned points.

See also

kdTreeNClosestPoint2D

Definition at line 549 of file KDTreeCapable.h.

kdTreeRadiusSearch2D()

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

size_t mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::kdTreeRadiusSearch2D ( const num_t  x0, const num_t  y0, const num_t  maxRadiusSqr, std::vector< std::pair< size_t, num_t >> &  out_indices_dist  ) const

inline

KD Tree-based search for all the points within a given radius of some 2D point.

This method automatically build the "m_kdtree_data" structure when:

• It is called for the first time
• The map has changed
• The KD-tree was build for 3D.

Parameters

x0	The X coordinate of the query.
y0	The Y coordinate of the query.
maxRadiusSqr	The square of the desired search radius.
out_indices_dist	The output list, with pairs of indeces/squared distances for the found correspondences.

Returns

Number of found points.

See also

kdTreeRadiusSearch3D, kdTreeNClosestPoint2DIdx

Definition at line 650 of file KDTreeCapable.h.

Referenced by ransac_data_assoc_run().

Here is the caller graph for this function:

kdTreeRadiusSearch3D()

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

size_t mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::kdTreeRadiusSearch3D ( const num_t  x0, const num_t  y0, const num_t  z0, const num_t  maxRadiusSqr, std::vector< std::pair< size_t, num_t >> &  out_indices_dist  ) const

inline

KD Tree-based search for all the points within a given radius of some 3D point.

This method automatically build the "m_kdtree_data" structure when:

• It is called for the first time
• The map has changed
• The KD-tree was build for 2D.

Parameters

x0	The X coordinate of the query.
y0	The Y coordinate of the query.
z0	The Z coordinate of the query.
maxRadiusSqr	The square of the desired search radius.
out_indices_dist	The output list, with pairs of indeces/squared distances for the found correspondences.

Returns

Number of found points.

See also

kdTreeRadiusSearch2D, kdTreeNClosestPoint3DIdx

Definition at line 615 of file KDTreeCapable.h.

kdTreeTwoClosestPoint2D() [1/2]

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

void mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::kdTreeTwoClosestPoint2D ( float  x0, float  y0, float &  out_x1, float &  out_y1, float &  out_x2, float &  out_y2, float &  out_dist_sqr1, float &  out_dist_sqr2  ) const

inline

KD Tree-based search for the TWO closest point to some given 2D coordinates.

This method automatically build the "m_kdtree_data" structure when:

• It is called for the first time
• The map has changed
• The KD-tree was build for 3D.

Parameters

x0	The X coordinate of the query.
y0	The Y coordinate of the query.
out_x1	The X coordinate of the first correspondence.
out_y1	The Y coordinate of the first correspondence.
out_x2	The X coordinate of the second correspondence.
out_y2	The Y coordinate of the second correspondence.
out_dist_sqr1	The square distance between the query and the first returned point.
out_dist_sqr2	The square distance between the query and the second returned point.

See also

kdTreeClosestPoint2D

Definition at line 233 of file KDTreeCapable.h.

Referenced by mrpt::math::KDTreeCapable< CFeatureList >::kdTreeTwoClosestPoint2D().

Here is the caller graph for this function:

kdTreeTwoClosestPoint2D() [2/2]

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

void mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::kdTreeTwoClosestPoint2D ( const TPoint2D &  p0, TPoint2D &  pOut1, TPoint2D &  pOut2, float &  outDistSqr1, float &  outDistSqr2  ) const

inline

Definition at line 264 of file KDTreeCapable.h.

operator=()

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

KDTreeCapable& mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::operator= ( const KDTreeCapable< Derived, num_t, metric_t > &  )

inline

Definition at line 93 of file KDTreeCapable.h.

rebuild_kdTree_2D()

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

void mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::rebuild_kdTree_2D ( ) const

inlineprivate

Rebuild, if needed the KD-tree for 2D (nDims=2), 3D (nDims=3), ...

asking the child class for the data points.

Definition at line 768 of file KDTreeCapable.h.

Referenced by mrpt::math::KDTreeCapable< CFeatureList >::kdTreeClosestPoint2D(), mrpt::math::KDTreeCapable< CFeatureList >::kdTreeEnsureIndexBuilt2D(), mrpt::math::KDTreeCapable< CFeatureList >::kdTreeNClosestPoint2D(), mrpt::math::KDTreeCapable< CFeatureList >::kdTreeNClosestPoint2DIdx(), mrpt::math::KDTreeCapable< CFeatureList >::kdTreeRadiusSearch2D(), and mrpt::math::KDTreeCapable< CFeatureList >::kdTreeTwoClosestPoint2D().

Here is the caller graph for this function:

rebuild_kdTree_3D()

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

void mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::rebuild_kdTree_3D ( ) const

inlineprivate

Rebuild, if needed the KD-tree for 2D (nDims=2), 3D (nDims=3), ...

asking the child class for the data points.

Definition at line 803 of file KDTreeCapable.h.

Referenced by mrpt::math::KDTreeCapable< CFeatureList >::kdTreeClosestPoint3D(), mrpt::math::KDTreeCapable< CFeatureList >::kdTreeEnsureIndexBuilt3D(), mrpt::math::KDTreeCapable< CFeatureList >::kdTreeNClosestPoint3D(), mrpt::math::KDTreeCapable< CFeatureList >::kdTreeNClosestPoint3DIdx(), mrpt::math::KDTreeCapable< CFeatureList >::kdTreeNClosestPoint3DWithIdx(), and mrpt::math::KDTreeCapable< CFeatureList >::kdTreeRadiusSearch3D().

Here is the caller graph for this function:

Member Data Documentation

kdtree_search_params

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

TKDTreeSearchParams mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::kdtree_search_params

Parameters to tune the ANN searches.

Definition at line 114 of file KDTreeCapable.h.

Referenced by mrpt::math::KDTreeCapable< CFeatureList >::rebuild_kdTree_2D(), and mrpt::math::KDTreeCapable< CFeatureList >::rebuild_kdTree_3D().

m_kdtree2d_data

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

TKDTreeDataHolder<2> mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::m_kdtree2d_data

mutableprivate

Definition at line 761 of file KDTreeCapable.h.

Referenced by mrpt::math::KDTreeCapable< CFeatureList >::kdTreeClosestPoint2D(), mrpt::math::KDTreeCapable< CFeatureList >::kdTreeNClosestPoint2D(), mrpt::math::KDTreeCapable< CFeatureList >::kdTreeNClosestPoint2DIdx(), mrpt::math::KDTreeCapable< CFeatureList >::kdTreeRadiusSearch2D(), mrpt::math::KDTreeCapable< CFeatureList >::kdTreeTwoClosestPoint2D(), mrpt::math::KDTreeCapable< CFeatureList >::rebuild_kdTree_2D(), and mrpt::math::KDTreeCapable< CFeatureList >::rebuild_kdTree_3D().

m_kdtree3d_data

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

TKDTreeDataHolder<3> mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::m_kdtree3d_data

mutableprivate

Definition at line 762 of file KDTreeCapable.h.

Referenced by mrpt::math::KDTreeCapable< CFeatureList >::kdTreeClosestPoint3D(), mrpt::math::KDTreeCapable< CFeatureList >::kdTreeNClosestPoint3D(), mrpt::math::KDTreeCapable< CFeatureList >::kdTreeNClosestPoint3DIdx(), mrpt::math::KDTreeCapable< CFeatureList >::kdTreeNClosestPoint3DWithIdx(), mrpt::math::KDTreeCapable< CFeatureList >::kdTreeRadiusSearch3D(), mrpt::math::KDTreeCapable< CFeatureList >::rebuild_kdTree_2D(), and mrpt::math::KDTreeCapable< CFeatureList >::rebuild_kdTree_3D().

m_kdtree_is_uptodate

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

std::atomic_bool mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::m_kdtree_is_uptodate {false}

mutableprivate

whether the KD tree needs to be rebuilt or not.

Definition at line 764 of file KDTreeCapable.h.

Referenced by mrpt::math::KDTreeCapable< CFeatureList >::kdtree_mark_as_outdated(), mrpt::math::KDTreeCapable< CFeatureList >::rebuild_kdTree_2D(), and mrpt::math::KDTreeCapable< CFeatureList >::rebuild_kdTree_3D().

m_kdtree_mtx

template<class Derived, typename num_t = float, typename metric_t = nanoflann::L2_Simple_Adaptor<num_t, Derived>>

std::mutex mrpt::math::KDTreeCapable< Derived, num_t, metric_t >::m_kdtree_mtx

mutableprivate

Definition at line 760 of file KDTreeCapable.h.

Referenced by mrpt::math::KDTreeCapable< CFeatureList >::kdtree_mark_as_outdated(), mrpt::math::KDTreeCapable< CFeatureList >::rebuild_kdTree_2D(), and mrpt::math::KDTreeCapable< CFeatureList >::rebuild_kdTree_3D().