boofcv.alg.tracker.sfot

Class SparseFlowObjectTracker<Image extends boofcv.struct.image.ImageGray<Image>,Derivative extends boofcv.struct.image.ImageGray<Derivative>>

java.lang.Object

boofcv.alg.tracker.sfot.SparseFlowObjectTracker<Image,Derivative>

public class SparseFlowObjectTracker<Image extends boofcv.struct.image.ImageGray<Image>,Derivative extends boofcv.struct.image.ImageGray<Derivative>>
extends java.lang.Object

Uses a pyramidal KLT tracker to track features inside the user selected region. The motion of the region is found robustly using LeastMedianOfSquares and a translation + rotation model. Drift is a problem since motion is estimated relative to the previous frame and it will eventually drift away from the original target. When it works well it is very smooth and can handle partially obscured objects. Can't recover after the target has been lost. Runs very fast.

Constructor Summary

Constructors

Constructor and Description
SparseFlowObjectTracker(SfotConfig config, java.lang.Class<Image> imageType, java.lang.Class<Derivative> derivType, boofcv.abst.filter.derivative.ImageGradient<Image,Derivative> gradient)

Method Summary

Modifier and Type	Method and Description
SfotConfig	getConfig()
void	init(Image input, boofcv.struct.RectangleRotate_F64 region)
boolean	isTrackLost()
boolean	update(Image input, boofcv.struct.RectangleRotate_F64 output) Given the input image compute the new location of the target region and store the results in output.

Methods inherited from class java.lang.Object

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

Constructor Detail

SparseFlowObjectTracker

public SparseFlowObjectTracker(SfotConfig config,
                               java.lang.Class<Image> imageType,
                               java.lang.Class<Derivative> derivType,
                               boofcv.abst.filter.derivative.ImageGradient<Image,Derivative> gradient)

Method Detail

init

public void init(Image input,
                 boofcv.struct.RectangleRotate_F64 region)

update

public boolean update(Image input,
                      boofcv.struct.RectangleRotate_F64 output)

Given the input image compute the new location of the target region and store the results in output.

Parameters:

input - next image in the sequence.

output - Storage for the output.

Returns:

true if tracking is successful

isTrackLost

public boolean isTrackLost()

getConfig

public SfotConfig getConfig()