boofcv.alg.tracker.tld

Class TldTemplateMatching<T extends boofcv.struct.image.ImageGray<T>>

java.lang.Object

boofcv.alg.tracker.tld.TldTemplateMatching<T>

public class TldTemplateMatching<T extends boofcv.struct.image.ImageGray<T>>
extends java.lang.Object

Created NCC templates to describe the target region. Each template is composed of a 15x15 area. The descriptor is computed by sampling evenly spaced points through out the rectangular region. Confidence values are computed based in the distance a point is from the closest positive and negative template.

Field Summary

Fields

Modifier and Type	Field and Description
protected java.util.Stack<boofcv.struct.feature.NccFeature>	unused

Constructor Summary

Constructors

Modifier	Constructor and Description
protected	TldTemplateMatching()
	TldTemplateMatching(boofcv.alg.interpolate.InterpolatePixelS<T> interpolate)

Method Summary

Modifier and Type	Method and Description
void	addDescriptor(boolean positive, float x0, float y0, float x1, float y1)
void	addDescriptor(boolean positive, boofcv.struct.ImageRectangle rect) Creates a new descriptor for the specified region
double	computeConfidence(boofcv.struct.ImageRectangle r) see the other function with the same name
double	computeConfidence(int x0, int y0, int x1, int y1) Compute a value which indicates how confident the specified region is to be a member of the positive set.
void	computeNccDescriptor(boofcv.struct.feature.NccFeature f, float x0, float y0, float x1, float y1) Computes the NCC descriptor by sample points at evenly spaced distances inside the rectangle
boofcv.struct.feature.NccFeature	createDescriptor() Creates a new descriptor or recycles an old one
double	distance(boofcv.struct.feature.NccFeature observed, java.util.List<boofcv.struct.feature.NccFeature> candidates) Computes the best distance to 'observed' from the candidate list.
java.util.List<boofcv.struct.feature.NccFeature>	getTemplateNegative()
java.util.List<boofcv.struct.feature.NccFeature>	getTemplatePositive()
void	reset() Discard previous results and puts it back into its initial state
void	setImage(T gray) Must call this function before any of the others which process descriptions

Methods inherited from class java.lang.Object

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

Field Detail

unused

protected java.util.Stack<boofcv.struct.feature.NccFeature> unused

Constructor Detail

TldTemplateMatching

public TldTemplateMatching(boofcv.alg.interpolate.InterpolatePixelS<T> interpolate)

TldTemplateMatching

protected TldTemplateMatching()

Method Detail

reset

public void reset()

Discard previous results and puts it back into its initial state

setImage

public void setImage(T gray)

Must call this function before any of the others which process descriptions

Parameters:

gray - Input image

addDescriptor

public void addDescriptor(boolean positive,
                          boofcv.struct.ImageRectangle rect)

Creates a new descriptor for the specified region

Parameters:

positive - if it is a positive or negative example

addDescriptor

public void addDescriptor(boolean positive,
                          float x0,
                          float y0,
                          float x1,
                          float y1)

computeNccDescriptor

public void computeNccDescriptor(boofcv.struct.feature.NccFeature f,
                                 float x0,
                                 float y0,
                                 float x1,
                                 float y1)

Computes the NCC descriptor by sample points at evenly spaced distances inside the rectangle

createDescriptor

public boofcv.struct.feature.NccFeature createDescriptor()

Creates a new descriptor or recycles an old one

computeConfidence

public double computeConfidence(int x0,
                                int y0,
                                int x1,
                                int y1)

Compute a value which indicates how confident the specified region is to be a member of the positive set. The confidence value is from 0 to 1. 1 indicates 100% confidence. Positive and negative templates are used to compute the confidence value. Only the point in each set which is closest to the specified region are used in the calculation.

Returns:

value from 0 to 1, where higher values are more confident

computeConfidence

public double computeConfidence(boofcv.struct.ImageRectangle r)

see the other function with the same name

distance

public double distance(boofcv.struct.feature.NccFeature observed,
                       java.util.List<boofcv.struct.feature.NccFeature> candidates)

Computes the best distance to 'observed' from the candidate list.

Parameters:

observed - Feature being matched

candidates - Set of candidate matches

Returns:

score from 0 to 1, where lower is closer

getTemplatePositive

public java.util.List<boofcv.struct.feature.NccFeature> getTemplatePositive()

getTemplateNegative

public java.util.List<boofcv.struct.feature.NccFeature> getTemplateNegative()