A wrapper around a KalmanFilter type object; additional logic can be provided through extension. More...

#include <TrackingFilter.hpp>

Inheritance diagram for TrackingFilter:

Collaboration diagram for TrackingFilter:

Public Member Functions
void	flushKalman ()
	Reset KF state to initial values.

double	kalmanDistance (cv::Mat measurement)
	Compute the Kalman distance to a measurement vector. More...

void	updateTrackFailure ()
	Alert the KF that we did not track a ball this frame.

cv::Mat	prediction ()
	The KF prediction of the state.

cv::Mat	covariance ()
	The KF's covariance matrix.

void	updateTimeState (double t)
	Update the time state of the tracker to the current time.

double	dT ()
	Change in time from prior state to this one.

double	time ()
	The current time state.

bool	isFound ()
	Whether we have found a object recently.

bool	isLost ()
	Whether the object is lost and we don't know anything about it.

Protected Attributes
int	mBlobRad
	Radius of the current state blob?

int	mLatency
	Unused?

int	mNotFoundCount
	Number of frames we have missed this object for. More...

int	mKfStateLen
	Length of the state vector; 6 <x,y,dx,dy,w,h>

cv::Mat	mKfState
	State vector of most recent state.

int	mFoundCount
	Number of frames we have tracked this object for.

int	mKfControlLen

cv::Mat	mKfControl

cv::Mat	mKfControlVec

int	mKfMeasLen
	Length of the measurement vector; 4 <x,y,w,h>

cv::Mat	mKfMeas
	Measurement vector of most recent state.

double	mTstart = 0
	Start time of this tick (the previous timestep's time)

double	mTstop = 0
	Stop time of this tick (the current timestate)

KalmanFilterPlus	mKf
	The Kalman Filter object proper.

Detailed Description

A wrapper around a KalmanFilter type object; additional logic can be provided through extension.

The idea is this is an "abstract" tracker which is object-agnostic; children should implement some methods with additional knowledge of the type of thing that they are trying to track.