Big and complicated class that implements the MHT (Multiple Hypotheses Tracking) algorithm. More...

#include <mht_declaration.h>

Public Member Functions
void	clear (void)

void	clearTargetAssociations (void)
	Go through all clusters, all hypotheses and finally all targets, removing target to measurement associations from past targets. More...

hypothesisTreePtr	getHypothesisTree (void)

double	getMaxGating (void)
	Return the maximum Mahalanobis gating distance. More...

string	getReportName (string extra)

vector< TargetPtr >	getTargets (void)

void	iterate (vector< MeasurementPtr > &measurements)
	Currently the main mht function that does all the work. More...

	Mht (void)
	Mht constructor. More...

void	simplifySingleTargetClusters (vector< MeasurementPtr > &global_measurements)

void	solveCluster (ClusterPtr cluster)

void	solveClusterMultiTread (ClusterPtr cluster)

void	solveHypothesis (HypothesisPtr parent_hypothesis, vector< MeasurementPtr > &measurements, vector< HypothesisPtr > &list_of_hypotheses)

	~Mht (void)
	Mht destructor. More...

Public Attributes
string	_aux1

vector< ClusterPtr >	_clusters
	Vector of current existing hypotheses clusters. More...

int	_j
	J parameter used in the propagation of hypotheses, must be configured form outside class. More...

int	_k
	K parameter used in the Murty's algorithm, must be configured form outside class. More...

Private Member Functions
TargetPtr	associate (TargetPtr &target, MeasurementPtr &measurement)

void	breakClusters (vector< MeasurementPtr > &measurements)

void	checkCurrentTargets (vector< MeasurementPtr > &measurements)

void	cleanClusters (void)
	Remove empty clusters. More...

void	clusterAssignemnt (vector< MeasurementPtr > &measurements)
	Distribute the measurements by the existing clusters. More...

void	clusterMerger (vector< MeasurementPtr > &measurements)
	Check if clusters need merging and do it!! More...

vector< ClusterPtr >::iterator	completeRemoveCluster (long id)
	Complete deletion of a cluster. More...

vector< ClusterPtr >::iterator	completeRemoveCluster (vector< ClusterPtr >::iterator it_cluster)
	Remove the cluster with a given id along with all the hypotheses associated with it. More...

bool	consistencyCheck (void)

void	createClusters (void)

vector< TargetPtr >	createCopy (vector< TargetPtr > &original)

ClusterPtr	findCluster (vector< ClusterPtr > &clusters, int id)

vector< HypothesisPtr >	findHypotheses (int cl, int it_past=1)
	Find all hypotheses with a given cluster id. More...

MeasurementPtr	findMeasurement (vector< MeasurementPtr > &measurements, long id)

HypothesisPtr	getHypothesisPtr (long id, ClusterPtr &cluster)

void	getProbability (HypothesisPtr hypothesis)

double	getProbability (uint n_det, uint n_occ, uint n_del, uint n_new, uint n_fal, double prod, double previous)

HypothesisTree::iterator	insertForcedParent (long cluster)

bool	normalize (vector< HypothesisPtr > &hypotheses, uint j=10e3)

void	recursiveHypothesisRemove (HypothesisTree::iterator it)
	Remove the hypothesis and possible parent if empty. More...

void	removeEmptyClusters (void)

void	removeIrrelevant (void)
	Remove branches that were not propagated and old hypotheses. More...

void	removeTheDead (void)
	Remove dead hypotheses. More...

void	switchColor (HypothesisPtr hypothesis)
	Change hypothesis color based on its current cluster. More...

void	updateTargetList (void)

Private Attributes
bool	_clean_old_modifying_tree
	Flag that controls the removal of old hypotheses, with changes to the tree structure. More...

long	_cluster_id
	Cluster id seed. More...

vector< TargetPtr >	_current_targets
	Vector to store outgoing targets. More...

uint	_history_size
	Number of previous state positions to store. More...

HypothesisTree::leaf_iterator	_hleaf
	Leaf auxiliary iterator. More...

hypothesisTreePtr	_hypothesisTree
	Main tree holding variable. More...

HypothesisTree::iterator	_iterator
	Simple iterator. More...

double	_max_gating
	Maximum gating threshold. More...

double	_minimum_representativity
	Minimum representativity value in propagating hypotheses. More...

int	_miss_association_threshold
	Maximum number of miss associations. More...

bool	_parent_tagging
	Flag that determines if the main parents are to be tagged, used in graphviz. More...

bool	_remove_empty_clusters
	Flag that controls the removal of empty clusters. More...

bool	_remove_unused
	Flag that determines if dead hypotheses are to be deleted, used in graphviz. More...

bool	_use_minimum_representativity
	Flag that controls the use of the minimum representativity rule. More...

long	iteration
	Current iteration. More...

Detailed Description

Big and complicated class that implements the MHT (Multiple Hypotheses Tracking) algorithm.

Create empty tree
Obtain new measurements

for each leaf
Gate measurements with tree leafs
Add gated measurements as leaf children
If no association was created iterate on void
Remove branch if needed
end for

for each measurement
if no associations
add new tree root node
end for

common iteration processing
tree pruning
nth pruning //delayed decision
age deletion //careful only works if age is bigger that nth pruning

Definition at line 147 of file mht_declaration.h.

Constructor & Destructor Documentation

Mht::Mht ( void )

Mht constructor.

Definition at line 185 of file mht_implementation.cpp.

Mht::~Mht ( void )

Mht destructor.

Definition at line 233 of file mht_implementation.cpp.

Member Function Documentation

TargetPtr Mht::associate	(	TargetPtr &	target,
		MeasurementPtr &	measurement
	)

private

Definition at line 2620 of file mht_implementation.cpp.

void Mht::breakClusters ( vector< MeasurementPtr > & measurements )

private

Definition at line 1986 of file mht_implementation.cpp.

void Mht::checkCurrentTargets ( vector< MeasurementPtr > & measurements )

private

Definition at line 2809 of file mht_implementation.cpp.

void Mht::cleanClusters ( void )

private

Remove empty clusters.

Definition at line 1864 of file mht_implementation.cpp.

void Mht::clear ( void )

Definition at line 1348 of file mht_implementation.cpp.

void Mht::clearTargetAssociations ( void )

Go through all clusters, all hypotheses and finally all targets, removing target to measurement associations from past targets.

Definition at line 242 of file mht_implementation.cpp.

void Mht::clusterAssignemnt ( vector< MeasurementPtr > & measurements )

private

Distribute the measurements by the existing clusters.

Definition at line 1812 of file mht_implementation.cpp.

void Mht::clusterMerger ( vector< MeasurementPtr > & measurements )

private

Check if clusters need merging and do it!!

This is quite complicated, merging involves a total reevaluation of targets, measurements and probabilities.

Definition at line 2377 of file mht_implementation.cpp.

vector< ClusterPtr >::iterator Mht::completeRemoveCluster ( long id )

private

Complete deletion of a cluster.

Parameters

id	of the cluster to remove

Returns: cluster vector iterator

Definition at line 1600 of file mht_implementation.cpp.

vector< ClusterPtr >::iterator Mht::completeRemoveCluster ( vector< ClusterPtr >::iterator it_cluster )

private

Remove the cluster with a given id along with all the hypotheses associated with it.

Definition at line 1613 of file mht_implementation.cpp.

bool Mht::consistencyCheck ( void )

private

Definition at line 1911 of file mht_implementation.cpp.

void Mht::createClusters ( void )

private

Definition at line 1760 of file mht_implementation.cpp.

vector< TargetPtr > Mht::createCopy ( vector< TargetPtr > & original )

private

Definition at line 2344 of file mht_implementation.cpp.

ClusterPtr Mht::findCluster	(	vector< ClusterPtr > &	clusters,
		int	id
	)

private

Definition at line 1749 of file mht_implementation.cpp.

vector< HypothesisPtr > Mht::findHypotheses	(	int	cl,
		int	it_past = `1`
	)

private

Find all hypotheses with a given cluster id.

This function will find all hypotheses that are currently leafs of the main tree with a given cluster. This function will create a std::vector of the HypothesisPtr and return it.

Warning: Only LEAF nodes will be returned! Not the whole branch!

Parameters

cl	cluster id
it_past	select the iteration from which the hypotheses will be extracted (default 1, previous iteration)

Returns: vector of hypothesisPtr

Definition at line 1722 of file mht_implementation.cpp.

MeasurementPtr Mht::findMeasurement	(	vector< MeasurementPtr > &	measurements,
		long	id
	)

private

Definition at line 1655 of file mht_implementation.cpp.

HypothesisPtr Mht::getHypothesisPtr	(	long	id,
		ClusterPtr &	cluster
	)

private

Definition at line 2733 of file mht_implementation.cpp.

hypothesisTreePtr Mht::getHypothesisTree ( void )

Definition at line 1337 of file mht_implementation.cpp.

double Mht::getMaxGating ( void )

Return the maximum Mahalanobis gating distance.

Definition at line 237 of file mht_implementation.cpp.

void Mht::getProbability ( HypothesisPtr hypothesis )

private

Definition at line 2686 of file mht_implementation.cpp.

double Mht::getProbability	(	uint	n_det,
		uint	n_occ,
		uint	n_del,
		uint	n_new,
		uint	n_fal,
		double	prod,
		double	previous
	)

private

Definition at line 2700 of file mht_implementation.cpp.

string Mht::getReportName ( string extra )

Definition at line 1360 of file mht_implementation.cpp.

vector< TargetPtr > Mht::getTargets ( void )

Definition at line 1342 of file mht_implementation.cpp.

HypothesisTree::iterator Mht::insertForcedParent ( long cluster )

private

Definition at line 2330 of file mht_implementation.cpp.

void Mht::iterate ( vector< MeasurementPtr > & measurements )

Currently the main mht function that does all the work.

This function starts by going through all the current leafs on the tree, for each leaf/hypothesis go through all targets and for each target check which measurements fall within its gate and add those measurements to the hypothesis cluster.

This is called measurement to cluster assignment, the end result is the cluster_assignemnt map.

The following step checks if measurements belong to more that one cluster and merges those clusters, if a measurement is shared this means that both clusters should be evaluated simultaneously. The merging of clusters is complicated: the first step is to define which cluster name will remain and which will disappear, I choose the first assigned to the measurement (for no particular reason). Then we go through all clusters associated with the measurement changing the associated hypothesis to the new cluster and all measurements associated with this cluster changing their association as well. In the end all clusters have different measurements without any overlap.

The next step is to create the list of problems to solve, each problem is a association of all the measurements of that cluster with the hypothesis that the cluster contains. This problems will be solved using the Murty algorithm with the Hungarian as a lower layer.

There are still a few steps missing ...

In the end, all measurements that were not associated with any cluster will create new clusters with a single hypothesis and a single target. This step makes the global tree grow.

Parameters

measurements input vector of measurements

Definition at line 1155 of file mht_implementation.cpp.

bool Mht::normalize	(	vector< HypothesisPtr > &	hypotheses,
		uint	j = `10e3`
	)

private

Definition at line 2665 of file mht_implementation.cpp.

void Mht::recursiveHypothesisRemove ( HypothesisTree::iterator it )

private

Remove the hypothesis and possible parent if empty.

Definition at line 1383 of file mht_implementation.cpp.

void Mht::removeEmptyClusters ( void )

private

Definition at line 1878 of file mht_implementation.cpp.

void Mht::removeIrrelevant ( void )

private

Remove branches that were not propagated and old hypotheses.

Definition at line 1401 of file mht_implementation.cpp.

void Mht::removeTheDead ( void )

private

Remove dead hypotheses.

Definition at line 1561 of file mht_implementation.cpp.

void Mht::simplifySingleTargetClusters ( vector< MeasurementPtr > & global_measurements )

Definition at line 262 of file mht_implementation.cpp.

void Mht::solveCluster ( ClusterPtr cluster )

Definition at line 724 of file mht_implementation.cpp.

void Mht::solveClusterMultiTread ( ClusterPtr cluster )

Definition at line 585 of file mht_implementation.cpp.

void Mht::solveHypothesis	(	HypothesisPtr	parent_hypothesis,
		vector< MeasurementPtr > &	measurements,
		vector< HypothesisPtr > &	list_of_hypotheses
	)

Definition at line 325 of file mht_implementation.cpp.

void Mht::switchColor ( HypothesisPtr hypothesis )

private

Change hypothesis color based on its current cluster.

This functions checks the cluster id of the hypothesis and calculates the corresponding color by doing cluster_id mod 8.
The result will be the color id.

Parameters

hypothesis the working node

Definition at line 1663 of file mht_implementation.cpp.

void Mht::updateTargetList ( void )

private

Definition at line 2741 of file mht_implementation.cpp.

Member Data Documentation

string Mht::_aux1

Definition at line 153 of file mht_declaration.h.

bool Mht::_clean_old_modifying_tree

private

Flag that controls the removal of old hypotheses, with changes to the tree structure.

Definition at line 255 of file mht_declaration.h.

long Mht::_cluster_id

private

Cluster id seed.

Definition at line 257 of file mht_declaration.h.

vector<ClusterPtr> Mht::_clusters

Vector of current existing hypotheses clusters.

Definition at line 152 of file mht_declaration.h.

vector<TargetPtr> Mht::_current_targets

private

Vector to store outgoing targets.

Definition at line 233 of file mht_declaration.h.

uint Mht::_history_size

private

Number of previous state positions to store.

Definition at line 245 of file mht_declaration.h.

HypothesisTree::leaf_iterator Mht::_hleaf

private

Leaf auxiliary iterator.

Definition at line 228 of file mht_declaration.h.

hypothesisTreePtr Mht::_hypothesisTree

private

Main tree holding variable.

Definition at line 225 of file mht_declaration.h.

HypothesisTree::iterator Mht::_iterator

private

Simple iterator.

Definition at line 230 of file mht_declaration.h.

int Mht::_j

J parameter used in the propagation of hypotheses, must be configured form outside class.

Definition at line 220 of file mht_declaration.h.

int Mht::_k

K parameter used in the Murty's algorithm, must be configured form outside class.

Definition at line 218 of file mht_declaration.h.

double Mht::_max_gating

private

Maximum gating threshold.

Definition at line 238 of file mht_declaration.h.

double Mht::_minimum_representativity

private

Minimum representativity value in propagating hypotheses.

Definition at line 242 of file mht_declaration.h.

int Mht::_miss_association_threshold

private

Maximum number of miss associations.

Definition at line 240 of file mht_declaration.h.

bool Mht::_parent_tagging

private

Flag that determines if the main parents are to be tagged, used in graphviz.

Definition at line 249 of file mht_declaration.h.

bool Mht::_remove_empty_clusters

private

Flag that controls the removal of empty clusters.

Definition at line 253 of file mht_declaration.h.

bool Mht::_remove_unused

private

Flag that determines if dead hypotheses are to be deleted, used in graphviz.

Definition at line 247 of file mht_declaration.h.

bool Mht::_use_minimum_representativity

private

Flag that controls the use of the minimum representativity rule.

Definition at line 251 of file mht_declaration.h.

long Mht::iteration

private

Current iteration.

Definition at line 236 of file mht_declaration.h.

The documentation for this class was generated from the following files:

Public Member Functions

Public Attributes

Private Member Functions

Private Attributes

Detailed Description

Constructor & Destructor Documentation

Member Function Documentation

Member Data Documentation