Package boofcv.alg.tracker.klt

Class PyramidKltTracker<InputImage extends ImageGray<InputImage>,DerivativeImage extends ImageGray<DerivativeImage>>

java.lang.Object

boofcv.alg.tracker.klt.PyramidKltTracker<InputImage,DerivativeImage>

public class PyramidKltTracker<InputImage extends ImageGray<InputImage>,DerivativeImage extends ImageGray<DerivativeImage>>
extends Object

A pyramid Kanade-Lucas-Tomasi (KLT) tracker that allows features to be tracker over a larger region than the basic (KltTracker) implementation. A feature is tracked at multiple resolutions, large motions can be detected at low resolution and are refined at higher resolutions.

Tracking is allowed along the image border. A track is dropped if the smallest track is outside the image.

Field Summary

Fields

Modifier and Type	Field	Description
protected @Nullable DerivativeImage[]	derivX
protected @Nullable DerivativeImage[]	derivY
protected ImagePyramid<InputImage>	image
protected KltTracker<InputImage,DerivativeImage>	tracker

Constructor Summary

Constructors

Constructor	Description
PyramidKltTracker(KltTracker<InputImage,DerivativeImage> tracker)

Method Summary

Modifier and Type	Method	Description
PyramidKltTracker<InputImage,DerivativeImage>	copyConcurrent()	Creates a copy which can be run in parallel.
float	getError()	Average error between track template and the image.
boolean	setDescription(PyramidKltFeature feature)	Sets the feature's description up.
void	setImage(ImagePyramid<InputImage> image)	Only sets the image pyramid.
void	setImage(ImagePyramid<InputImage> image, DerivativeImage[] derivX, DerivativeImage[] derivY)	Sets the current input images for the tracker to use.
KltTrackFault	track(PyramidKltFeature feature)	Finds the feature's new location in the image.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Details

tracker

protected KltTracker<InputImage extends ImageGray<InputImage>,DerivativeImage extends ImageGray<DerivativeImage>> tracker

image

protected ImagePyramid<InputImage extends ImageGray<InputImage>> image

derivX

@Nullable
protected @Nullable DerivativeImage extends ImageGray<DerivativeImage>[] derivX

derivY

@Nullable
protected @Nullable DerivativeImage extends ImageGray<DerivativeImage>[] derivY

Constructor Details

PyramidKltTracker

public PyramidKltTracker(KltTracker<InputImage,DerivativeImage> tracker)

Method Details

setImage

public void setImage(ImagePyramid<InputImage> image,
 DerivativeImage[] derivX,
 DerivativeImage[] derivY)

Sets the current input images for the tracker to use.

Parameters:

image - Original image pyramid.

derivX - Derivative along x-axis.

derivY - Derivative along y-axis.

setImage

public void setImage(ImagePyramid<InputImage> image)

Only sets the image pyramid. The derivatives are set to null. Only use this when tracking.

Parameters:

image - Image pyramid

setDescription

public boolean setDescription(PyramidKltFeature feature)

Sets the feature's description up. The feature's (x,y) must have already been set and setImage(boofcv.struct.pyramid.ImagePyramid<InputImage>, DerivativeImage[], DerivativeImage[]) been called.

Parameters:

feature - Feature's whose description is being setup.

Returns:

true if there was sufficient information to create a feature or false if not

track

public KltTrackFault track(PyramidKltFeature feature)

Finds the feature's new location in the image. The feature's position can be modified even if tracking fails.

NOTE: The feature's description is not updated and tracking over several frames can break down if its description is not updated.

Parameters:

feature - The feature being tracked.

Returns:

If tracking failed or not.

getError

public float getError()

Average error between track template and the image.

Returns:

error

See Also:

• boofcv.alg.tracker.klt.KltTracker#getError()

copyConcurrent

public PyramidKltTracker<InputImage,DerivativeImage> copyConcurrent()

Creates a copy which can be run in parallel. Data structures which can be shared are shared.