boofcv.abst.tracker

Class ConfigComaniciu2003

java.lang.Object

boofcv.abst.tracker.ConfigComaniciu2003

public class ConfigComaniciu2003
extends java.lang.Object

Configuration for Comaniciu2003_to_TrackerObjectQuad.

Field Summary

Fields

Modifier and Type	Field and Description
TypeInterpolate	interpolation Which interpolation method should it use.
float	maxPixelValue Largest value a pixel can have.
int	meanShiftMaxIterations Maximum number of mean-shift iterations.
float	meanShiftMinimumChange Mean-shift will stop when the change is below this threshold.
float	minimumSizeRatio The scale is allowed to be reduced by this much from the original region which is selected.
int	numHistogramBins Number of histogram bins for each band.
int	numSamples Number of points it samples along each axis of the rectangle.
double	numSigmas Used to compute weights.
float	scaleChange Specifies how much it will scale the region up and down by when testing for a scale change.
float	scaleWeight Weighting factor which limits the amount it will change the scale.
boolean	updateHistogram If true the histogram will be updated using the most recent image.

Constructor Summary

Constructors

Constructor and Description
ConfigComaniciu2003()
ConfigComaniciu2003(boolean estimateScale)
ConfigComaniciu2003(int numSamples, int numHistogramBins, float scaleWeight)

Method Summary

Methods inherited from class java.lang.Object

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

Field Detail

numSamples

public int numSamples

Number of points it samples along each axis of the rectangle. Default is 30.

numSigmas

public double numSigmas

Used to compute weights. Number of standard deviations away the sides will be from the center. Shouldn't need to tune this. Try 3.

numHistogramBins

public int numHistogramBins

Number of histogram bins for each band. Try 5

maxPixelValue

public float maxPixelValue

Largest value a pixel can have. For 8-bit images this is 255. Floating point images are some times normalized to 1.

updateHistogram

public boolean updateHistogram

If true the histogram will be updated using the most recent image. Try false.

meanShiftMaxIterations

public int meanShiftMaxIterations

Maximum number of mean-shift iterations. Try 30

meanShiftMinimumChange

public float meanShiftMinimumChange

Mean-shift will stop when the change is below this threshold. Try 1e-4f

scaleWeight

public float scaleWeight

Weighting factor which limits the amount it will change the scale. Value from 0 to 1. Closer to 0 the more it will prefer the most recent estimate. Try 0.1

scaleChange

public float scaleChange

Specifies how much it will scale the region up and down by when testing for a scale change. Allowed values are from 0 to 1, inclusive. 0 means no scale change and 1 is 100% increase and decrease. If no scale change is considered it can run 3x faster. If the target doesn't change scale then the tracker is much more robust. The paper recommends 0.1. By default scale change is set to 0.

minimumSizeRatio

public float minimumSizeRatio

The scale is allowed to be reduced by this much from the original region which is selected. Default is 0.25

interpolation

public TypeInterpolate interpolation

Which interpolation method should it use.

Constructor Detail

ConfigComaniciu2003

public ConfigComaniciu2003(int numSamples,
                           int numHistogramBins,
                           float scaleWeight)

ConfigComaniciu2003

public ConfigComaniciu2003(boolean estimateScale)

ConfigComaniciu2003

public ConfigComaniciu2003()