jafar::lines::LsPredictor2D Class Reference

Class for prediction of the movement of a line in a sequenze of images. More...

---

Detailed Description

Class for prediction of the movement of a line in a sequenze of images.

This Class provides containers and functions to predict the new position of a line in an image after it was tracked over prior images.

Definition at line 40 of file lsPredictor2D.hpp.

#include <lsPredictor2D.hpp>

List of all members.

Public Member Functions
void	initPredictor (double mx, double my, double phi, double rho, double theta, PredictionModel model=PERP_MIDPOINT_PROJ)
	Initializes the Kalman filters for prediction.
void	updatePredictor (double mxOld, double myOld, double phiOld, double mxNew, double myNew, double phiNew, double rhoNew, double thetaNew)
	Updates the predition model.
double	getPredictionEP (double x1Old, double y1Old, double x2Old, double y2Old, double &x1Pred, double &y1Pred, double &x2Pred, double &y2Pred, double l=0)
	Get the prediction for new endpoints.
double	benchmarkTest ()
	Get the benchmark value.
Private Attributes
PredictionModel	predMod
	Type of motion model.
uint	nUpdates
	Number of updates.
double	cDist
	Distance of last prediction to measurement.
double	benchmark
	Indicator for the reliability of the prediction, small value is good.
filter::ConstantVelocityKalmanFilter	xKF
	Constant velocity modell for x coordinate of midpoint.
filter::ConstantVelocityKalmanFilter	yKF
	Constant velocity modell for y coordinate of midpoint.
filter::ConstantVelocityKalmanFilter	phiKF
	Constant velocity modell for direction of line.
filter::ConstantVelocityKalmanFilter	rhoKF
	Constant velocity modell for polar coordinate rho.
filter::ConstantVelocityKalmanFilter	thetaKF
	Constant velocity modell for polar coordinate theta.
filter::ConstantVelocityKalmanFilter	dKF
	Constant position modell for perpendicular difference of lines.
filter::ConstantVelocityKalmanFilter	alphaKF
	Constant velocity modell for angular difference.
filter::ConstantVelocityKalmanFilter	xPKF
	Constant velocity modell for x coordinate of midpoint, updated with perpendicular motion.
filter::ConstantVelocityKalmanFilter	yPKF
	Constant velocity modell for y coordinate of midpoint, updated with perpendicular motion.
filter::ConstantVelocityKalmanFilter	phiPKF
	Constant velocity modell for direction of line for perpendicular update.

---

Member Function Documentation

double jafar::lines::LsPredictor2D::benchmarkTest

(

)

[inline]

Get the benchmark value.

This function provides a value that indicates if we can trust in the prediction modell. That test is not well realised for all prediction models. Should be improved.

Definition at line 98 of file lsPredictor2D.hpp.

References benchmark.

double jafar::lines::LsPredictor2D::getPredictionEP	(	double	x1Old,
		double	y1Old,
		double	x2Old,
		double	y2Old,
		double &	x1Pred,
		double &	y1Pred,
		double &	x2Pred,
		double &	y2Pred,
		double	l = 0
	)

Get the prediction for new endpoints.

This function provides the prediction of a line in endpoint representation. The motion prediction is only reliable if the line was tracked over prior states and if the mortion did not change to much. A constant velocity model is used.

void jafar::lines::LsPredictor2D::initPredictor	(	double	mx,
		double	my,
		double	phi,
		double	rho,
		double	theta,
		PredictionModel	model = PERP_MIDPOINT_PROJ
	)

Initializes the Kalman filters for prediction.

Initializes the Kalman filters for prediction. The recommended prediction model is PERP_MIDPOINT_PROJ. Depending on the prediction model the parameters that are really necessary changes.

MIDPOINT_X_Y Requires mx, my, phi

POLAR_COORD, Requires rho, theta

ONLY_PERP_MOTION, Requires mx my phi

PERP_MIDPOINT_PROJ Requires mx, my, phi

void jafar::lines::LsPredictor2D::updatePredictor	(	double	mxOld,
		double	myOld,
		double	phiOld,
		double	mxNew,
		double	myNew,
		double	phiNew,
		double	rhoNew,
		double	thetaNew
	)

Updates the predition model.

Updates the Kalman filters for prediction. The variances for measurements and process noise are hard encoded. Depending on the prediction model the parameters that are really necessary changes.

MIDPOINT_X_Y Requires mxNew, myNew, phiNew

POLAR_COORD Requires rhoNew, thetaNew

ONLY_PERP_MOTION Requires mxOld myOld phiOld mxNew myNew phiNew

PERP_MIDPOINT_PROJ Requires mxNew, myNew, phiNew

---

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

• modules/lines/include/lines/lsPredictor2D.hpp