A mixture of Gaussians adaptive updater based on D. More...

#include <vpdt_update_mog.h>

Inheritance diagram for vpdt_mog_lee_updater< mog_type >:

Public Types
typedef mog_type::component_type	gaussian_type

typedef gaussian_type::field_type	F

typedef vpdt_field_traits< F >::scalar_type	T

typedef mog_type	distribution_type

typedef mog_type::field_type	field_type

Public Member Functions
	vpdt_mog_lee_updater (const gaussian_type &init_gaussian, unsigned int max_cmp=5, T g_thresh=T(2.5), T min_stdev=T(0.16), bool use_wta=false)
	Constructor. More...

void	operator() (mog_type &mix, const F &sample) const
	The main function. More...

Protected Member Functions
T	total_num_obs (const mog_type &mix) const
	Count the total number of observations in all components. More...

std::vector< std::pair< unsigned int, double > >	matches (const mog_type &mix, const F &sample) const
	find all matches within the `gt2_` threshold and compute the probability of each. More...

void	winner_take_all (std::vector< std::pair< unsigned int, double > > &m) const
	Apply a winner-take-all strategy to the matches. More...

void	update (mog_type &mix, const F &sample, T alpha) const
	Update the mixture with sample using learning rate alpha. More...

void	insert (mog_type &mixture, const F &sample, T init_weight) const
	insert a sample in the mixture. More...

unsigned int	match (const mog_type &mix, const F &sample, const T &gt2, T &sqr_dist) const
	Return the index of the first Gaussian within the threshold distance. More...

Protected Attributes
T	gt2_
	Squared Gaussian Mahalanobis distance threshold. More...

T	min_var_
	Minimum variance allowed in each Gaussian component. More...

bool	use_winner_take_all_
	Use a winner-take_all strategy. More...

gaussian_type	init_gaussian_
	A model for new Gaussians inserted. More...

unsigned int	max_components_
	The maximum number of components in the mixture. More...

Detailed Description

template<class mog_type>
class vpdt_mog_lee_updater< mog_type >

A mixture of Gaussians adaptive updater based on D.

-S. Lee. Modification of the S-G method. Each Gaussian has its own learning rate that adjusts with the number of observations. This algorithm requires that the Gaussians in the mixture are wrapped in a vpdt_num_obs class to count the number of observations. This algorithm is based on: D.-S. Lee. "Effective gaussian mixture learning for video background subtraction", PAMI, 27:827–832, May 2005.

Definition at line 170 of file vpdt_update_mog.h.

Member Typedef Documentation

◆ distribution_type

template<class mog_type >

typedef mog_type vpdt_mog_updater< mog_type >::distribution_type

inherited

Definition at line 28 of file vpdt_update_mog.h.

◆ F

template<class mog_type >

typedef gaussian_type::field_type vpdt_mog_lee_updater< mog_type >::F

Definition at line 174 of file vpdt_update_mog.h.

◆ field_type

template<class mog_type >

typedef mog_type::field_type vpdt_mog_updater< mog_type >::field_type

inherited

Definition at line 29 of file vpdt_update_mog.h.

◆ gaussian_type

template<class mog_type >

typedef mog_type::component_type vpdt_mog_lee_updater< mog_type >::gaussian_type

Definition at line 173 of file vpdt_update_mog.h.

◆ T

template<class mog_type >

typedef vpdt_field_traits<F>::scalar_type vpdt_mog_lee_updater< mog_type >::T

Definition at line 175 of file vpdt_update_mog.h.

Constructor & Destructor Documentation

◆ vpdt_mog_lee_updater()

template<class mog_type >

vpdt_mog_lee_updater< mog_type >::vpdt_mog_lee_updater	(	const gaussian_type &	init_gaussian,
		unsigned int	max_cmp = `5`,
		T	g_thresh = `T(2.5)`,
		T	min_stdev = `T(0.16)`,
		bool	use_wta = `false`
	)

inline

Constructor.

Definition at line 178 of file vpdt_update_mog.h.

Member Function Documentation

◆ insert()

template<class mog_type >

void vpdt_mog_updater< mog_type >::insert	(	mog_type &	mixture,
		const F &	sample,
		T	init_weight
	)		const

inlineprotectedinherited

insert a sample in the mixture.

Parameters

sample	A Gaussian is inserted with a mean of sample and a covariance from init_gaussian_
init_weight	The normalized weight the resulting sample should have after insertion

Definition at line 47 of file vpdt_update_mog.h.

◆ match()

template<class mog_type >

unsigned int vpdt_mog_updater< mog_type >::match	(	const mog_type &	mix,
		const F &	sample,
		const T &	gt2,
		T &	sqr_dist
	)		const

inlineprotectedinherited

Return the index of the first Gaussian within the threshold distance.

The threshold gt2 is on the square distance. The computed square distance is returned by reference in sqr_dist If there are no matches return the number of components (last index + 1)

Definition at line 73 of file vpdt_update_mog.h.

◆ matches()

template<class mog_type >

std::vector<std::pair<unsigned int,double> > vpdt_mog_lee_updater< mog_type >::matches	(	const mog_type &	mix,
		const F &	sample
	)		const

inlineprotected

find all matches within the gt2_ threshold and compute the probability of each.

Definition at line 210 of file vpdt_update_mog.h.

◆ operator()()

template<class mog_type >

void vpdt_mog_lee_updater< mog_type >::operator()	(	mog_type &	mix,
		const F &	sample
	)		const

inline

The main function.

Definition at line 188 of file vpdt_update_mog.h.

◆ total_num_obs()

template<class mog_type >

T vpdt_mog_lee_updater< mog_type >::total_num_obs ( const mog_type & mix ) const

inlineprotected

Count the total number of observations in all components.

Definition at line 197 of file vpdt_update_mog.h.

◆ update()

template<class mog_type >

void vpdt_mog_lee_updater< mog_type >::update	(	mog_type &	mix,
		const F &	sample,
		T	alpha
	)		const

inlineprotected

Update the mixture with sample using learning rate alpha.

Definition at line 261 of file vpdt_update_mog.h.

◆ winner_take_all()

template<class mog_type >

void vpdt_mog_lee_updater< mog_type >::winner_take_all ( std::vector< std::pair< unsigned int, double > > & m ) const

inlineprotected

Apply a winner-take-all strategy to the matches.

Keep only the highest probability match and assign it probability 1

Definition at line 245 of file vpdt_update_mog.h.

Member Data Documentation

◆ gt2_

template<class mog_type >

T vpdt_mog_lee_updater< mog_type >::gt2_

protected

Squared Gaussian Mahalanobis distance threshold.

Definition at line 300 of file vpdt_update_mog.h.

◆ init_gaussian_

template<class mog_type >

gaussian_type vpdt_mog_updater< mog_type >::init_gaussian_

mutableprotectedinherited

A model for new Gaussians inserted.

Definition at line 87 of file vpdt_update_mog.h.

◆ max_components_

template<class mog_type >

unsigned int vpdt_mog_updater< mog_type >::max_components_

protectedinherited

The maximum number of components in the mixture.

Definition at line 89 of file vpdt_update_mog.h.

◆ min_var_

template<class mog_type >

T vpdt_mog_lee_updater< mog_type >::min_var_

protected

Minimum variance allowed in each Gaussian component.

Definition at line 302 of file vpdt_update_mog.h.

◆ use_winner_take_all_

template<class mog_type >

bool vpdt_mog_lee_updater< mog_type >::use_winner_take_all_

protected

Use a winner-take_all strategy.

Definition at line 305 of file vpdt_update_mog.h.

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

core/vpdl/vpdt/vpdt_update_mog.h

Public Types

Public Member Functions

Protected Member Functions

Protected Attributes

Detailed Description

template<class mog_type> class vpdt_mog_lee_updater< mog_type >

Member Typedef Documentation

◆ distribution_type

◆ F

◆ field_type

◆ gaussian_type

◆ T

Constructor & Destructor Documentation

◆ vpdt_mog_lee_updater()

Member Function Documentation

◆ insert()

◆ match()

◆ matches()

◆ operator()()

◆ total_num_obs()

◆ update()

◆ winner_take_all()

Member Data Documentation

◆ gt2_

◆ init_gaussian_

◆ max_components_

◆ min_var_

◆ use_winner_take_all_

template<class mog_type>
class vpdt_mog_lee_updater< mog_type >