#include <kmeans_index.h>

Inheritance diagram for rtflann::KMeansIndex< Distance >:

Classes
struct	Node
struct	PointInfo
Public Types
typedef NNIndex< Distance >	BaseClass
typedef Distance::ResultType	DistanceType
typedef Distance::ElementType	ElementType
typedef bool	needs_vector_space_distance
Public Member Functions
void	addPoints (const Matrix< ElementType > &points, float rebuild_threshold=2)
	Incrementally add points to the index.
BaseClass *	clone () const
void	findNeighbors (ResultSet< DistanceType > &result, const ElementType *vec, const SearchParams &searchParams) const
int	getClusterCenters (Matrix< DistanceType > &centers)
flann_algorithm_t	getType () const
void	initCenterChooser ()
	KMeansIndex (const Matrix< ElementType > &inputData, const IndexParams &params=KMeansIndexParams(), Distance d=Distance())
	KMeansIndex (const IndexParams &params=KMeansIndexParams(), Distance d=Distance())
	KMeansIndex (const KMeansIndex &other)
void	loadIndex (FILE *stream)
KMeansIndex &	operator= (KMeansIndex other)
void	saveIndex (FILE *stream)
template<typename Archive >
void	serialize (Archive &ar)
void	set_cb_index (float index)
int	usedMemory () const
virtual	~KMeansIndex ()
Protected Member Functions
void	buildIndexImpl ()
Private Types
typedef BranchStruct< NodePtr, DistanceType >	BranchSt
typedef Node *	NodePtr
Private Member Functions
void	addPointToTree (NodePtr node, size_t index, DistanceType dist_to_pivot)
void	computeClustering (NodePtr node, int *indices, int indices_length, int branching)
void	computeNodeStatistics (NodePtr node, const std::vector< int > &indices)
void	copyTree (NodePtr &dst, const NodePtr &src)
int	exploreNodeBranches (NodePtr node, const ElementType q, Heap< BranchSt > heap) const
template<bool with_removed>
void	findExactNN (NodePtr node, ResultSet< DistanceType > &result, const ElementType *vec) const
template<bool with_removed>
void	findNeighborsWithRemoved (ResultSet< DistanceType > &result, const ElementType *vec, const SearchParams &searchParams) const
template<bool with_removed>
void	findNN (NodePtr node, ResultSet< DistanceType > &result, const ElementType vec, int &checks, int maxChecks, Heap< BranchSt > heap) const
void	freeIndex ()
void	getCenterOrdering (NodePtr node, const ElementType *q, std::vector< int > &sort_indices) const
DistanceType	getDistanceToBorder (DistanceType p, DistanceType c, DistanceType *q) const
int	getMinVarianceClusters (NodePtr root, std::vector< NodePtr > &clusters, int clusters_length, DistanceType &varianceValue) const
void	swap (KMeansIndex &other)
Private Attributes
int	branching_
float	cb_index_
flann_centers_init_t	centers_init_
CenterChooser< Distance > *	chooseCenters_
int	iterations_
int	memoryCounter_
PooledAllocator	pool_
NodePtr	root_

Detailed Description

template<typename Distance>
class rtflann::KMeansIndex< Distance >

Hierarchical kmeans index

Contains a tree constructed through a hierarchical kmeans clustering and other information for indexing a set of points for nearest-neighbour matching.

Definition at line 83 of file kmeans_index.h.

Member Typedef Documentation

template<typename Distance>

typedef NNIndex<Distance> rtflann::KMeansIndex< Distance >::BaseClass

Definition at line 89 of file kmeans_index.h.

template<typename Distance>

typedef BranchStruct<NodePtr, DistanceType> rtflann::KMeansIndex< Distance >::BranchSt [private]

Alias definition for a nicer syntax.

Definition at line 454 of file kmeans_index.h.

template<typename Distance>

typedef Distance::ResultType rtflann::KMeansIndex< Distance >::DistanceType

Reimplemented from rtflann::NNIndex< Distance >.

Definition at line 87 of file kmeans_index.h.

template<typename Distance>

typedef Distance::ElementType rtflann::KMeansIndex< Distance >::ElementType

Reimplemented from rtflann::NNIndex< Distance >.

Definition at line 86 of file kmeans_index.h.

template<typename Distance>

typedef bool rtflann::KMeansIndex< Distance >::needs_vector_space_distance

Definition at line 91 of file kmeans_index.h.

template<typename Distance>

typedef Node* rtflann::KMeansIndex< Distance >::NodePtr [private]

Definition at line 449 of file kmeans_index.h.

Constructor & Destructor Documentation

template<typename Distance>

rtflann::KMeansIndex< Distance >::KMeansIndex	(	const Matrix< ElementType > &	inputData,
		const IndexParams &	params = `KMeansIndexParams()`,
		Distance	d = `Distance()`
	)		`[inline]`

Index constructor

Params: inputData = dataset with the input features params = parameters passed to the hierarchical k-means algorithm

Definition at line 107 of file kmeans_index.h.

template<typename Distance>

rtflann::KMeansIndex< Distance >::KMeansIndex	(	const IndexParams &	params = `KMeansIndexParams()`,
		Distance	d = `Distance()`
	)		`[inline]`

Index constructor

Params: inputData = dataset with the input features params = parameters passed to the hierarchical k-means algorithm

Definition at line 131 of file kmeans_index.h.

template<typename Distance>

rtflann::KMeansIndex< Distance >::KMeansIndex ( const KMeansIndex< Distance > & other ) [inline]

Definition at line 146 of file kmeans_index.h.

template<typename Distance>

virtual rtflann::KMeansIndex< Distance >::~KMeansIndex ( ) [inline, virtual]

Index destructor.

Release the memory used by the index.

Definition at line 187 of file kmeans_index.h.

Member Function Documentation

template<typename Distance>

void rtflann::KMeansIndex< Distance >::addPoints	(	const Matrix< ElementType > &	points,
		float	rebuild_threshold = `2`
	)		`[inline, virtual]`

Incrementally add points to the index.

Parameters:

points	Matrix with points to be added
rebuild_threshold

Reimplemented from rtflann::NNIndex< Distance >.

Definition at line 215 of file kmeans_index.h.

template<typename Distance>

void rtflann::KMeansIndex< Distance >::addPointToTree	(	NodePtr	node,
		size_t	index,
		DistanceType	dist_to_pivot
	)		`[inline, private]`

Definition at line 969 of file kmeans_index.h.

template<typename Distance>

void rtflann::KMeansIndex< Distance >::buildIndexImpl ( ) [inline, protected, virtual]

Builds the index

Implements rtflann::NNIndex< Distance >.

Definition at line 329 of file kmeans_index.h.

template<typename Distance>

BaseClass* rtflann::KMeansIndex< Distance >::clone ( ) const [inline, virtual]

Implements rtflann::NNIndex< Distance >.

Definition at line 193 of file kmeans_index.h.

template<typename Distance>

void rtflann::KMeansIndex< Distance >::computeClustering	(	NodePtr	node,
		int *	indices,
		int	indices_length,
		int	branching
	)		`[inline, private]`

The method responsible with actually doing the recursive hierarchical clustering

Params: node = the node to cluster indices = indices of the points belonging to the current node branching = the branching factor to use in the clustering

TODO: for 1-sized clusters don't store a cluster center (it's the same as the single cluster point)

Definition at line 542 of file kmeans_index.h.

template<typename Distance>

void rtflann::KMeansIndex< Distance >::computeNodeStatistics	(	NodePtr	node,
		const std::vector< int > &	indices
	)		`[inline, private]`

Computes the statistics of a node (mean, radius, variance).

Params: node = the node to use indices = the indices of the points belonging to the node

Definition at line 495 of file kmeans_index.h.

template<typename Distance>

void rtflann::KMeansIndex< Distance >::copyTree	(	NodePtr &	dst,
		const NodePtr &	src
	)		`[inline, private]`

Definition at line 467 of file kmeans_index.h.

template<typename Distance>

int rtflann::KMeansIndex< Distance >::exploreNodeBranches	(	NodePtr	node,
		const ElementType *	q,
		Heap< BranchSt > *	heap
	)		const `[inline, private]`

Helper function that computes the nearest childs of a node to a given query point. Params: node = the node q = the query point distances = array with the distances to each child node. Returns:

Definition at line 802 of file kmeans_index.h.

template<typename Distance>

template<bool with_removed>

void rtflann::KMeansIndex< Distance >::findExactNN	(	NodePtr	node,
		ResultSet< DistanceType > &	result,
		const ElementType *	vec
	)		const `[inline, private]`

Function the performs exact nearest neighbor search by traversing the entire tree.

Definition at line 835 of file kmeans_index.h.

template<typename Distance>

void rtflann::KMeansIndex< Distance >::findNeighbors	(	ResultSet< DistanceType > &	result,
		const ElementType *	vec,
		const SearchParams &	searchParams
	)		const `[inline, virtual]`

Find set of nearest neighbors to vec. Their indices are stored inside the result object.

Params: result = the result object in which the indices of the nearest-neighbors are stored vec = the vector for which to search the nearest neighbors searchParams = parameters that influence the search algorithm (checks, cb_index)

Implements rtflann::NNIndex< Distance >.

Definition at line 283 of file kmeans_index.h.

template<typename Distance>

template<bool with_removed>

void rtflann::KMeansIndex< Distance >::findNeighborsWithRemoved	(	ResultSet< DistanceType > &	result,
		const ElementType *	vec,
		const SearchParams &	searchParams
	)		const `[inline, private]`

Definition at line 715 of file kmeans_index.h.

template<typename Distance>

template<bool with_removed>

void rtflann::KMeansIndex< Distance >::findNN	(	NodePtr	node,
		ResultSet< DistanceType > &	result,
		const ElementType *	vec,
		int &	checks,
		int	maxChecks,
		Heap< BranchSt > *	heap
	)		const `[inline, private]`

Performs one descent in the hierarchical k-means tree. The branches not visited are stored in a priority queue.

Params: node = node to explore result = container for the k-nearest neighbors found vec = query points checks = how many points in the dataset have been checked so far maxChecks = maximum dataset points to checks

Definition at line 755 of file kmeans_index.h.

template<typename Distance>

void rtflann::KMeansIndex< Distance >::freeIndex ( ) [inline, private, virtual]

Helper function

Implements rtflann::NNIndex< Distance >.

Definition at line 460 of file kmeans_index.h.

template<typename Distance>

void rtflann::KMeansIndex< Distance >::getCenterOrdering	(	NodePtr	node,
		const ElementType *	q,
		std::vector< int > &	sort_indices
	)		const `[inline, private]`

Helper function.

I computes the order in which to traverse the child nodes of a particular node.

Definition at line 880 of file kmeans_index.h.

template<typename Distance>

int rtflann::KMeansIndex< Distance >::getClusterCenters ( Matrix< DistanceType > & centers ) [inline]

Clustering function that takes a cut in the hierarchical k-means tree and return the clusters centers of that clustering. Params: numClusters = number of clusters to have in the clustering computed Returns: number of cluster centers

Definition at line 301 of file kmeans_index.h.

template<typename Distance>

DistanceType rtflann::KMeansIndex< Distance >::getDistanceToBorder	(	DistanceType *	p,
		DistanceType *	c,
		DistanceType *	q
	)		const `[inline, private]`

Method that computes the squared distance from the query point q from inside region with center c to the border between this region and the region with center p

Definition at line 902 of file kmeans_index.h.

template<typename Distance>

int rtflann::KMeansIndex< Distance >::getMinVarianceClusters	(	NodePtr	root,
		std::vector< NodePtr > &	clusters,
		int	clusters_length,
		DistanceType &	varianceValue
	)		const `[inline, private]`

Helper function the descends in the hierarchical k-means tree by spliting those clusters that minimize the overall variance of the clustering. Params: root = root node clusters = array with clusters centers (return value) varianceValue = variance of the clustering (return value) Returns:

Definition at line 926 of file kmeans_index.h.

template<typename Distance>

flann_algorithm_t rtflann::KMeansIndex< Distance >::getType ( ) const [inline, virtual]

Implements rtflann::IndexBase.

Definition at line 95 of file kmeans_index.h.

template<typename Distance>

void rtflann::KMeansIndex< Distance >::initCenterChooser ( ) [inline]

Definition at line 165 of file kmeans_index.h.

template<typename Distance>

void rtflann::KMeansIndex< Distance >::loadIndex ( FILE * stream ) [inline, virtual]

Implements rtflann::IndexBase.

Definition at line 266 of file kmeans_index.h.

template<typename Distance>

KMeansIndex& rtflann::KMeansIndex< Distance >::operator= ( KMeansIndex< Distance > other ) [inline]

Definition at line 158 of file kmeans_index.h.

template<typename Distance>

void rtflann::KMeansIndex< Distance >::saveIndex ( FILE * stream ) [inline, virtual]

Implements rtflann::IndexBase.

Definition at line 260 of file kmeans_index.h.

template<typename Distance>

template<typename Archive >

void rtflann::KMeansIndex< Distance >::serialize ( Archive & ar ) [inline]

Reimplemented from rtflann::NNIndex< Distance >.

Definition at line 234 of file kmeans_index.h.

template<typename Distance>

void rtflann::KMeansIndex< Distance >::set_cb_index ( float index ) [inline]

Definition at line 199 of file kmeans_index.h.

template<typename Distance>

void rtflann::KMeansIndex< Distance >::swap ( KMeansIndex< Distance > & other ) [inline, private]

Definition at line 1010 of file kmeans_index.h.

template<typename Distance>

int rtflann::KMeansIndex< Distance >::usedMemory ( ) const [inline, virtual]

Computes the inde memory usage Returns: memory used by the index

Implements rtflann::IndexBase.

Definition at line 208 of file kmeans_index.h.

Member Data Documentation

template<typename Distance>

int rtflann::KMeansIndex< Distance >::branching_ [private]

The branching factor used in the hierarchical k-means clustering

Definition at line 1025 of file kmeans_index.h.

template<typename Distance>

float rtflann::KMeansIndex< Distance >::cb_index_ [private]

Cluster border index. This is used in the tree search phase when determining the closest cluster to explore next. A zero value takes into account only the cluster centres, a value greater then zero also take into account the size of the cluster.

Definition at line 1039 of file kmeans_index.h.

template<typename Distance>

flann_centers_init_t rtflann::KMeansIndex< Distance >::centers_init_ [private]

Algorithm for choosing the cluster centers

Definition at line 1031 of file kmeans_index.h.

template<typename Distance>

CenterChooser<Distance>* rtflann::KMeansIndex< Distance >::chooseCenters_ [private]

Algorithm used to choose initial centers

Definition at line 1059 of file kmeans_index.h.

template<typename Distance>

int rtflann::KMeansIndex< Distance >::iterations_ [private]

Maximum number of iterations to use when performing k-means clustering