jafar::klt::Klt Class Reference

This class is a Jafar wrapper (using Image) to the KLT tracker. More...

---

Detailed Description

This class is a Jafar wrapper (using Image) to the KLT tracker.

This class wraps all functions of the klt.h header, which contains all necessary functions to use Birchfield's KLT implementation. Some functions have been added to offer a superior level of use of the tracker. Only added functions are documented, for others please refer to the KLT lib documentation (http://www.ces.clemson.edu/~stb/klt/). All methods have the same name than the original function but with the "KLT" prefix removed.

Warning:

All formats of Image will functions, but you should use u8 grey images with dimensions multiple of 8, because it will save some conversions with memory copies of images.

Definition at line 126 of file klt.hpp.

#include <klt.hpp>

List of all members.

Public Member Functions
	Klt (int maxDisplacement=50, int maxAccel=50, int securityMargin=10, int measureConfidence=10, int initSpeedConfidence=100)
	constructor
KLT_TrackingContext	CreateTrackingContext (void)
KLT_FeatureList	CreateFeatureList (int nFeatures)
KLT_FeatureHistory	CreateFeatureHistory (int nFrames)
KLT_FeatureTable	CreateFeatureTable (int nFrames, int nFeatures)
void	FreeTrackingContext (KLT_TrackingContext tc)
void	FreeFeatureList (KLT_FeatureList fl)
void	FreeFeatureHistory (KLT_FeatureHistory fh)
void	FreeFeatureTable (KLT_FeatureTable ft)
void	SelectGoodFeatures (KLT_TrackingContext tc, image::Image &jImg, KLT_FeatureList fl)
void	TrackFeatures (KLT_TrackingContext tc, image::Image &jImg1, image::Image &jImg2, KLT_FeatureList fl)
void	ReplaceLostFeatures (KLT_TrackingContext tc, image::Image &jImg, int ncols, int nrows, KLT_FeatureList fl)
void	SelectGoodFeaturesObject (KLT_TrackingContext tc, image::Image &jImg, KLT_FeatureList fl, int zone_x=-1, int zone_y=-1, int zone_w=-1, int zone_h=-1, tDate timeStamp=-1)
	select features in an image, restricted to a given area
void	TrackFeaturesObject (KLT_TrackingContext tc, image::Image &jImg1, image::Image &jImg2, KLT_FeatureList fl, bool replace=false, tDate timeStamp=-1)
	track features of the object from one image to the next, and replace features on the object if asked
int	CountRemainingFeatures (KLT_FeatureList fl)
void	PrintTrackingContext (KLT_TrackingContext tc)
void	ChangeTCPyramid (KLT_TrackingContext tc, int search_range)
void	UpdateTCBorder (KLT_TrackingContext tc)
void	StopSequentialMode (KLT_TrackingContext tc)
void	SetVerbosity (int verbosity)
float	ComputeSmoothSigma (KLT_TrackingContext tc)
void	StoreFeatureList (KLT_FeatureList fl, KLT_FeatureTable ft, int frame)
void	ExtractFeatureList (KLT_FeatureList fl, KLT_FeatureTable ft, int frame)
void	StoreFeatureHistory (KLT_FeatureHistory fh, KLT_FeatureTable ft, int feat)
void	ExtractFeatureHistory (KLT_FeatureHistory fh, KLT_FeatureTable ft, int feat)
void	WriteFeatureListToPPM (KLT_FeatureList fl, KLT_PixelType *greyimg, int ncols, int nrows, char *filename)
void	WriteImageToImg (image::Image &jGreyimg, image::Image &jRGBimg)
	copy grey image in rgb image
void	WriteFeatureListToImg (KLT_FeatureList fl, image::Image &jRGBimg)
	draw a representation of feature on the image
void	WriteBordersToImg (image::Image &jRGBimg, KLT_TrackingContext tc)
	draw a representation of the working area on the image
void	WriteKalmanToImg (image::Image &jRGBimg, KLT_TrackingContext tc)
	draw a representation of kalman informations on the image (positions, speeds, and variances)
void	WriteFeatureList (KLT_FeatureList fl, char *filename, char *fmt)
void	WriteFeatureHistory (KLT_FeatureHistory fh, char *filename, char *fmt)
void	WriteFeatureTable (KLT_FeatureTable ft, char *filename, char *fmt)
KLT_FeatureList	ReadFeatureList (KLT_FeatureList fl, char *filename)
KLT_FeatureHistory	ReadFeatureHistory (KLT_FeatureHistory fh, char *filename)
KLT_FeatureTable	ReadFeatureTable (KLT_FeatureTable ft, char *filename)
Public Attributes
TrackedObject	object
Private Attributes
int	workingPos [2]
int	prevWorkingPos [2]
int	workingDims [2]
int	prevWorkingDims [2]
int	searchingZone [4]
int	securityMargin
filter::KalmanFilter *	kalman
tDate	lastTimeStamp
	kalman filter (see filter jafar module)

---

Constructor & Destructor Documentation

jafar::klt::Klt::Klt	(	int	maxDisplacement = 50,
		int	maxAccel = 50,
		int	securityMargin = 10,
		int	measureConfidence = 10,
		int	initSpeedConfidence = 100
	)		[inline]

constructor

Parameters:

maxDisplacement	the maximum displacement in pixels from one frame to the next, determining the size of the processing area
maxAccel	the maximum acceleration of the object in pixels by frames^2, determining the error of the kalman model (speed is constant)
securityMargin	a margin to increase the size of the processing area
measureConfidence	the confidence in the measure (barycenter of features representing the object)
initSpeedConfidence	the confidence in the initial speed, that's to say the maximum value of initial speed (because we assume it to be 0 for the first frame)

Definition at line 165 of file klt.hpp.

---

Member Function Documentation

void jafar::klt::Klt::SelectGoodFeaturesObject	(	KLT_TrackingContext	tc,
		image::Image &	jImg,
		KLT_FeatureList	fl,
		int	zone_x = -1,
		int	zone_y = -1,
		int	zone_w = -1,
		int	zone_h = -1,
		tDate	timeStamp = -1
	)

select features in an image, restricted to a given area

This function is an enhancement of the KLT lib's function SelectGoodFeatures, enabling to restrict the selection of features to a given area of the image, or more concretely to a given object.

Parameters:

tc	the tracking context (please refer to the KLT lib documentation)
jImg	the image where select features
fl	the feature list for storing found features (please refer to the KLT lib documentation)
zone_x	the x of top left corner of the area where you want to select features in the first image
zone_y	the y of top left corner of the area where you want to select features in the first image
zone_w	the width of the area where you want to select features in the first image
zone_h	the height of the area where you want to select features in the first image
timeStamp	a time stamp of the image, if default (-1) time stamp is automatically increased of 1 at each frame, so that all speed units are in pixels /frame. If this field is used, to set different dt between frames, be carefull to set speeds in the appropriate unit

void jafar::klt::Klt::TrackFeaturesObject	(	KLT_TrackingContext	tc,
		image::Image &	jImg1,
		image::Image &	jImg2,
		KLT_FeatureList	fl,
		bool	replace = false,
		tDate	timeStamp = -1
	)

track features of the object from one image to the next, and replace features on the object if asked

This function is an enhancement of the KLT lib's functions TrackGoodFeatures and ReplaceLostFeatures, enabling to track features only in a small area (determined by a kalman filter) in order to speed up the processing, and to replace lost features on the object (with an elimination of features not on the object)

Parameters:

tc	the tracking context (please refer to the KLT lib documentation)
jImg1	the previous image. If sequentialMode is set, this image is generally not used, except if the size of the processing area is changing. So you should always give a correct parameter here.
jImg2	the new image where to track features
fl	the feature list (please refer to the KLT lib documentation)
replace	specify whether lost features should be replaced when lost or not
timeStamp	cf SelectGoodFeaturesObject

void jafar::klt::Klt::WriteBordersToImg	(	image::Image &	jRGBimg,
		KLT_TrackingContext	tc
	)

draw a representation of the working area on the image

Parameters:

jRGBimg	the destination image, which must be a RGB 3-channel image with same dimensions ncols and nrows than jGreyimage, and which MUST be already allocated !
tc	the tracking context used

void jafar::klt::Klt::WriteFeatureListToImg	(	KLT_FeatureList	fl,
		image::Image &	jRGBimg
	)

draw a representation of feature on the image

Parameters:

fl	the feature list to draw
jRGBimg	the destination image, which must be a RGB 3-channel image with same dimensions ncols and nrows than jGreyimage, and which MUST be already allocated !

void jafar::klt::Klt::WriteImageToImg	(	image::Image &	jGreyimg,
		image::Image &	jRGBimg
	)

copy grey image in rgb image

Parameters:

jGreyimage	the original grey image
jRGBimg	the destination image, which must be a RGB 3-channel image with same dimensions than jGreyimage, and which MUST be already allocated !

void jafar::klt::Klt::WriteKalmanToImg	(	image::Image &	jRGBimg,
		KLT_TrackingContext	tc
	)

draw a representation of kalman informations on the image (positions, speeds, and variances)

Parameters:

jRGBimg	the destination image, which must be a RGB 3-channel image with same dimensions ncols and nrows than jGreyimage, and which MUST be already allocated !
tc	the tracking context used

---

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

• modules/klt/include/klt/klt.hpp