boofcv.factory.feature.tracker

Class FactoryPointTracker

java.lang.Object

boofcv.factory.feature.tracker.FactoryPointTracker

public class FactoryPointTracker
extends java.lang.Object

Factory for creating trackers which implement PointTracker. These trackers are intended for use in SFM applications. Some features which individual trackers can provide are lost when using the high level interface PointTracker. To create low level tracking algorithms see FactoryTrackerAlg

See Also:

FactoryTrackerAlg

Constructor Summary

Constructors

Constructor and Description
FactoryPointTracker()

Method Summary

Modifier and Type	Method and Description
static <I extends boofcv.struct.image.ImageSingleBand> PointTracker<I>	combined_FH_SURF_KLT(boofcv.alg.tracker.klt.PkltConfig kltConfig, int reactivateThreshold, boofcv.abst.feature.detect.interest.ConfigFastHessian configDetector, boofcv.abst.feature.describe.ConfigSurfDescribe.Stability configDescribe, boofcv.abst.feature.orientation.ConfigSlidingIntegral configOrientation, java.lang.Class<I> imageType) Creates a tracker which detects Fast-Hessian features, describes them with SURF, nominally tracks them using KLT.
static <I extends boofcv.struct.image.ImageSingleBand,D extends boofcv.struct.image.ImageSingleBand> PointTracker<I>	combined_ST_SURF_KLT(boofcv.abst.feature.detect.interest.ConfigGeneralDetector configExtract, boofcv.alg.tracker.klt.PkltConfig kltConfig, int reactivateThreshold, boofcv.abst.feature.describe.ConfigSurfDescribe.Stability configDescribe, boofcv.abst.feature.orientation.ConfigSlidingIntegral configOrientation, java.lang.Class<I> imageType, java.lang.Class<D> derivType) Creates a tracker which detects Shi-Tomasi corner features, describes them with SURF, and nominally tracks them using KLT.
static <I extends boofcv.struct.image.ImageSingleBand,D extends boofcv.struct.image.ImageSingleBand,Desc extends boofcv.struct.feature.TupleDesc> PointTracker<I>	combined(boofcv.abst.feature.detdesc.DetectDescribePoint<I,Desc> detector, boofcv.abst.feature.associate.AssociateDescription<Desc> associate, boofcv.alg.tracker.klt.PkltConfig kltConfig, int reactivateThreshold, java.lang.Class<I> imageType) Creates a tracker that is a hybrid between KLT and Detect-Describe-Associate (DDA) trackers.
static <I extends boofcv.struct.image.ImageSingleBand,Desc extends boofcv.struct.feature.TupleDesc> PointTracker<I>	combined(boofcv.abst.feature.detect.interest.InterestPointDetector<I> detector, boofcv.abst.feature.orientation.OrientationImage<I> orientation, boofcv.abst.feature.describe.DescribeRegionPoint<I,Desc> describe, boofcv.abst.feature.associate.AssociateDescription<Desc> associate, boofcv.alg.tracker.klt.PkltConfig kltConfig, int reactivateThreshold, java.lang.Class<I> imageType) Creates a tracker that is a hybrid between KLT and Detect-Describe-Associate (DDA) trackers.
static <I extends boofcv.struct.image.ImageSingleBand,D extends boofcv.struct.image.ImageSingleBand> boofcv.alg.feature.detect.interest.GeneralFeatureDetector<I,D>	createShiTomasi(boofcv.abst.feature.detect.interest.ConfigGeneralDetector config, java.lang.Class<D> derivType) Creates a Shi-Tomasi corner detector specifically designed for SFM.
static <I extends boofcv.struct.image.ImageSingleBand,D extends boofcv.struct.image.ImageSingleBand> PointTracker<I>	dda_FAST_BRIEF(boofcv.abst.feature.detect.interest.ConfigFast configFast, boofcv.abst.feature.detect.interest.ConfigGeneralDetector configExtract, int maxAssociationError, java.lang.Class<I> imageType) Creates a tracker which detects FAST corner features and describes them with BRIEF.
static <I extends boofcv.struct.image.ImageSingleBand> PointTracker<I>	dda_FH_SURF_Fast(boofcv.abst.feature.detect.interest.ConfigFastHessian configDetector, boofcv.abst.feature.describe.ConfigSurfDescribe.Speed configDescribe, boofcv.abst.feature.orientation.ConfigAverageIntegral configOrientation, java.lang.Class<I> imageType) Creates a tracker which detects Fast-Hessian features and describes them with SURF using the faster variant of SURF.
static <I extends boofcv.struct.image.ImageSingleBand> PointTracker<I>	dda_FH_SURF_Stable(boofcv.abst.feature.detect.interest.ConfigFastHessian configDetector, boofcv.abst.feature.describe.ConfigSurfDescribe.Stability configDescribe, boofcv.abst.feature.orientation.ConfigSlidingIntegral configOrientation, java.lang.Class<I> imageType) Creates a tracker which detects Fast-Hessian features and describes them with SURF using the faster variant of SURF.
static <I extends boofcv.struct.image.ImageSingleBand,D extends boofcv.struct.image.ImageSingleBand> PointTracker<I>	dda_ST_BRIEF(int maxAssociationError, boofcv.abst.feature.detect.interest.ConfigGeneralDetector configExtract, java.lang.Class<I> imageType, java.lang.Class<D> derivType) Creates a tracker which detects Shi-Tomasi corner features and describes them with BRIEF.
static <I extends boofcv.struct.image.ImageSingleBand,D extends boofcv.struct.image.ImageSingleBand> PointTracker<I>	dda_ST_NCC(boofcv.abst.feature.detect.interest.ConfigGeneralDetector configExtract, int describeRadius, java.lang.Class<I> imageType, java.lang.Class<D> derivType) Creates a tracker which detects Shi-Tomasi corner features and describes them with NCC.
static <I extends boofcv.struct.image.ImageSingleBand,Desc extends boofcv.struct.feature.TupleDesc> DetectDescribeAssociate<I,Desc>	dda(boofcv.abst.feature.detdesc.DetectDescribePoint<I,Desc> detDesc, boofcv.abst.feature.associate.AssociateDescription2D<Desc> associate, boolean updateDescription)
static <I extends boofcv.struct.image.ImageSingleBand,D extends boofcv.struct.image.ImageSingleBand,Desc extends boofcv.struct.feature.TupleDesc> PointTracker<I>	dda(boofcv.alg.feature.detect.interest.GeneralFeatureDetector<I,D> detector, boofcv.abst.feature.describe.DescribeRegionPoint<I,Desc> describe, boofcv.abst.feature.associate.AssociateDescription2D<Desc> associate, double scale, java.lang.Class<I> imageType)
static <I extends boofcv.struct.image.ImageSingleBand,Desc extends boofcv.struct.feature.TupleDesc> DetectDescribeAssociate<I,Desc>	dda(boofcv.abst.feature.detect.interest.InterestPointDetector<I> detector, boofcv.abst.feature.orientation.OrientationImage<I> orientation, boofcv.abst.feature.describe.DescribeRegionPoint<I,Desc> describe, boofcv.abst.feature.associate.AssociateDescription2D<Desc> associate, boolean updateDescription) Creates a tracker which uses the detect, describe, associate architecture.
static <I extends boofcv.struct.image.ImageSingleBand,D extends boofcv.struct.image.ImageSingleBand> PointTracker<I>	klt(int[] scaling, boofcv.abst.feature.detect.interest.ConfigGeneralDetector configExtract, int featureRadius, java.lang.Class<I> imageType, java.lang.Class<D> derivType) Pyramid KLT feature tracker.
static <I extends boofcv.struct.image.ImageSingleBand,D extends boofcv.struct.image.ImageSingleBand> PointTracker<I>	klt(boofcv.alg.tracker.klt.PkltConfig config, boofcv.abst.feature.detect.interest.ConfigGeneralDetector configExtract, java.lang.Class<I> imageType, java.lang.Class<D> derivType) Pyramid KLT feature tracker.

Methods inherited from class java.lang.Object

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

Constructor Detail

FactoryPointTracker

public FactoryPointTracker()

Method Detail

klt

public static <I extends boofcv.struct.image.ImageSingleBand,D extends boofcv.struct.image.ImageSingleBand> PointTracker<I> klt(int[] scaling,
                                                                                                                                boofcv.abst.feature.detect.interest.ConfigGeneralDetector configExtract,
                                                                                                                                int featureRadius,
                                                                                                                                java.lang.Class<I> imageType,
                                                                                                                                java.lang.Class<D> derivType)

Pyramid KLT feature tracker.

Parameters:

scaling - Scales in the image pyramid. Recommend [1,2,4] or [2,4]

configExtract - Configuration for extracting features

featureRadius - Size of the tracked feature. Try 3 or 5

imageType - Input image type.

derivType - Image derivative type.

Returns:

KLT based tracker.

See Also:

PyramidKltTracker

klt

public static <I extends boofcv.struct.image.ImageSingleBand,D extends boofcv.struct.image.ImageSingleBand> PointTracker<I> klt(boofcv.alg.tracker.klt.PkltConfig config,
                                                                                                                                boofcv.abst.feature.detect.interest.ConfigGeneralDetector configExtract,
                                                                                                                                java.lang.Class<I> imageType,
                                                                                                                                java.lang.Class<D> derivType)

Pyramid KLT feature tracker.

Parameters:

config - Config for the tracker. Try PkltConfig.createDefault().

configExtract - Configuration for extracting features

Returns:

KLT based tracker.

See Also:

PyramidKltTracker

dda_FH_SURF_Fast

public static <I extends boofcv.struct.image.ImageSingleBand> PointTracker<I> dda_FH_SURF_Fast(boofcv.abst.feature.detect.interest.ConfigFastHessian configDetector,
                                                                                               boofcv.abst.feature.describe.ConfigSurfDescribe.Speed configDescribe,
                                                                                               boofcv.abst.feature.orientation.ConfigAverageIntegral configOrientation,
                                                                                               java.lang.Class<I> imageType)

Creates a tracker which detects Fast-Hessian features and describes them with SURF using the faster variant of SURF.

Parameters:

configDetector - Configuration for SURF detector

configDescribe - Configuration for SURF descriptor

configOrientation - Configuration for orientation

imageType - Type of image the input is.

Returns:

SURF based tracker.

See Also:

DescribePointSurf, DdaManagerDetectDescribePoint

dda_FH_SURF_Stable

public static <I extends boofcv.struct.image.ImageSingleBand> PointTracker<I> dda_FH_SURF_Stable(boofcv.abst.feature.detect.interest.ConfigFastHessian configDetector,
                                                                                                 boofcv.abst.feature.describe.ConfigSurfDescribe.Stability configDescribe,
                                                                                                 boofcv.abst.feature.orientation.ConfigSlidingIntegral configOrientation,
                                                                                                 java.lang.Class<I> imageType)

Creates a tracker which detects Fast-Hessian features and describes them with SURF using the faster variant of SURF.

Parameters:

configDetector - Configuration for SURF detector

configDescribe - Configuration for SURF descriptor

configOrientation - Configuration for orientation

imageType - Type of image the input is.

Returns:

SURF based tracker.

See Also:

DescribePointSurf, DdaManagerDetectDescribePoint

dda_ST_BRIEF

public static <I extends boofcv.struct.image.ImageSingleBand,D extends boofcv.struct.image.ImageSingleBand> PointTracker<I> dda_ST_BRIEF(int maxAssociationError,
                                                                                                                                         boofcv.abst.feature.detect.interest.ConfigGeneralDetector configExtract,
                                                                                                                                         java.lang.Class<I> imageType,
                                                                                                                                         java.lang.Class<D> derivType)

Creates a tracker which detects Shi-Tomasi corner features and describes them with BRIEF.

Parameters:

maxAssociationError - Maximum allowed association error. Try 200.

configExtract - Configuration for extracting features

imageType - Type of image being processed.

derivType - Type of image used to store the image derivative. null == use default

See Also:

ShiTomasiCornerIntensity, DescribePointBrief, DdaManagerDetectDescribePoint

dda_FAST_BRIEF

public static <I extends boofcv.struct.image.ImageSingleBand,D extends boofcv.struct.image.ImageSingleBand> PointTracker<I> dda_FAST_BRIEF(boofcv.abst.feature.detect.interest.ConfigFast configFast,
                                                                                                                                           boofcv.abst.feature.detect.interest.ConfigGeneralDetector configExtract,
                                                                                                                                           int maxAssociationError,
                                                                                                                                           java.lang.Class<I> imageType)

Creates a tracker which detects FAST corner features and describes them with BRIEF.

Parameters:

configFast - Configuration for FAST detector

configExtract - Configuration for extracting features

maxAssociationError - Maximum allowed association error. Try 200.

imageType - Type of image being processed.

See Also:

FastCornerIntensity, DescribePointBrief, DdaManagerDetectDescribePoint

dda_ST_NCC

public static <I extends boofcv.struct.image.ImageSingleBand,D extends boofcv.struct.image.ImageSingleBand> PointTracker<I> dda_ST_NCC(boofcv.abst.feature.detect.interest.ConfigGeneralDetector configExtract,
                                                                                                                                       int describeRadius,
                                                                                                                                       java.lang.Class<I> imageType,
                                                                                                                                       java.lang.Class<D> derivType)

Creates a tracker which detects Shi-Tomasi corner features and describes them with NCC.

Parameters:

configExtract - Configuration for extracting features

describeRadius - Radius of the region being described. Try 2.

imageType - Type of image being processed.

derivType - Type of image used to store the image derivative. null == use default

See Also:

ShiTomasiCornerIntensity, DescribePointPixelRegionNCC, DdaManagerDetectDescribePoint

dda

public static <I extends boofcv.struct.image.ImageSingleBand,Desc extends boofcv.struct.feature.TupleDesc> DetectDescribeAssociate<I,Desc> dda(boofcv.abst.feature.detect.interest.InterestPointDetector<I> detector,
                                                                                                                                               boofcv.abst.feature.orientation.OrientationImage<I> orientation,
                                                                                                                                               boofcv.abst.feature.describe.DescribeRegionPoint<I,Desc> describe,
                                                                                                                                               boofcv.abst.feature.associate.AssociateDescription2D<Desc> associate,
                                                                                                                                               boolean updateDescription)

Creates a tracker which uses the detect, describe, associate architecture.

Type Parameters:

I - Type of input image.

Desc - Type of region description

Parameters:

detector - Interest point detector.

orientation - Optional orientation estimation algorithm. Can be null.

describe - Region description.

associate - Description association.

updateDescription - After a track has been associated should the description be changed? Try false.

Returns:

tracker

dda

public static <I extends boofcv.struct.image.ImageSingleBand,Desc extends boofcv.struct.feature.TupleDesc> DetectDescribeAssociate<I,Desc> dda(boofcv.abst.feature.detdesc.DetectDescribePoint<I,Desc> detDesc,
                                                                                                                                               boofcv.abst.feature.associate.AssociateDescription2D<Desc> associate,
                                                                                                                                               boolean updateDescription)

combined_FH_SURF_KLT

public static <I extends boofcv.struct.image.ImageSingleBand> PointTracker<I> combined_FH_SURF_KLT(boofcv.alg.tracker.klt.PkltConfig kltConfig,
                                                                                                   int reactivateThreshold,
                                                                                                   boofcv.abst.feature.detect.interest.ConfigFastHessian configDetector,
                                                                                                   boofcv.abst.feature.describe.ConfigSurfDescribe.Stability configDescribe,
                                                                                                   boofcv.abst.feature.orientation.ConfigSlidingIntegral configOrientation,
                                                                                                   java.lang.Class<I> imageType)

Creates a tracker which detects Fast-Hessian features, describes them with SURF, nominally tracks them using KLT.

Type Parameters:

I - Input image type.

Parameters:

kltConfig - Configuration for KLT tracker

reactivateThreshold - Tracks are reactivated after this many have been dropped. Try 10% of maxMatches

configDetector - Configuration for SURF detector

configDescribe - Configuration for SURF descriptor

configOrientation - Configuration for region orientation

imageType - Type of image the input is.

Returns:

SURF based tracker.

See Also:

DescribePointSurf, DdaManagerDetectDescribePoint

combined_ST_SURF_KLT

public static <I extends boofcv.struct.image.ImageSingleBand,D extends boofcv.struct.image.ImageSingleBand> PointTracker<I> combined_ST_SURF_KLT(boofcv.abst.feature.detect.interest.ConfigGeneralDetector configExtract,
                                                                                                                                                 boofcv.alg.tracker.klt.PkltConfig kltConfig,
                                                                                                                                                 int reactivateThreshold,
                                                                                                                                                 boofcv.abst.feature.describe.ConfigSurfDescribe.Stability configDescribe,
                                                                                                                                                 boofcv.abst.feature.orientation.ConfigSlidingIntegral configOrientation,
                                                                                                                                                 java.lang.Class<I> imageType,
                                                                                                                                                 java.lang.Class<D> derivType)

Creates a tracker which detects Shi-Tomasi corner features, describes them with SURF, and nominally tracks them using KLT.

Parameters:

configExtract - Configuration for extracting features

kltConfig - Configuration for KLT

reactivateThreshold - Tracks are reactivated after this many have been dropped. Try 10% of maxMatches

configDescribe - Configuration for SURF descriptor

configOrientation - Configuration for region orientation. If null then orientation isn't estimated

imageType - Type of image the input is.

derivType - Image derivative type. @return SURF based tracker.

See Also:

ShiTomasiCornerIntensity, DescribePointSurf, DdaManagerDetectDescribePoint

combined

public static <I extends boofcv.struct.image.ImageSingleBand,Desc extends boofcv.struct.feature.TupleDesc> PointTracker<I> combined(boofcv.abst.feature.detect.interest.InterestPointDetector<I> detector,
                                                                                                                                    boofcv.abst.feature.orientation.OrientationImage<I> orientation,
                                                                                                                                    boofcv.abst.feature.describe.DescribeRegionPoint<I,Desc> describe,
                                                                                                                                    boofcv.abst.feature.associate.AssociateDescription<Desc> associate,
                                                                                                                                    boofcv.alg.tracker.klt.PkltConfig kltConfig,
                                                                                                                                    int reactivateThreshold,
                                                                                                                                    java.lang.Class<I> imageType)

Creates a tracker that is a hybrid between KLT and Detect-Describe-Associate (DDA) trackers.

Parameters:

detector - Feature detector.

orientation - Optional feature orientation. Can be null.

describe - Feature description

associate - Association algorithm.

kltConfig - Configuration for KLT tracker

reactivateThreshold - Tracks are reactivated after this many have been dropped. Try 10% of maxMatches

imageType - Input image type. @return Feature tracker

See Also:

CombinedTrackerScalePoint

combined

public static <I extends boofcv.struct.image.ImageSingleBand,D extends boofcv.struct.image.ImageSingleBand,Desc extends boofcv.struct.feature.TupleDesc> PointTracker<I> combined(boofcv.abst.feature.detdesc.DetectDescribePoint<I,Desc> detector,
                                                                                                                                                                                  boofcv.abst.feature.associate.AssociateDescription<Desc> associate,
                                                                                                                                                                                  boofcv.alg.tracker.klt.PkltConfig kltConfig,
                                                                                                                                                                                  int reactivateThreshold,
                                                                                                                                                                                  java.lang.Class<I> imageType)

Creates a tracker that is a hybrid between KLT and Detect-Describe-Associate (DDA) trackers.

Parameters:

detector - Feature detector and describer.

associate - Association algorithm.

kltConfig - Configuration for KLT tracker

reactivateThreshold - Tracks are reactivated after this many have been dropped. Try 10% of maxMatches

imageType - Input image type. @return Feature tracker

See Also:

CombinedTrackerScalePoint

dda

public static <I extends boofcv.struct.image.ImageSingleBand,D extends boofcv.struct.image.ImageSingleBand,Desc extends boofcv.struct.feature.TupleDesc> PointTracker<I> dda(boofcv.alg.feature.detect.interest.GeneralFeatureDetector<I,D> detector,
                                                                                                                                                                             boofcv.abst.feature.describe.DescribeRegionPoint<I,Desc> describe,
                                                                                                                                                                             boofcv.abst.feature.associate.AssociateDescription2D<Desc> associate,
                                                                                                                                                                             double scale,
                                                                                                                                                                             java.lang.Class<I> imageType)

createShiTomasi

public static <I extends boofcv.struct.image.ImageSingleBand,D extends boofcv.struct.image.ImageSingleBand> boofcv.alg.feature.detect.interest.GeneralFeatureDetector<I,D> createShiTomasi(boofcv.abst.feature.detect.interest.ConfigGeneralDetector config,
                                                                                                                                                                                           java.lang.Class<D> derivType)

Creates a Shi-Tomasi corner detector specifically designed for SFM. Smaller feature radius work better. Variable detectRadius to control the number of features. When larger features are used weighting should be set to true, but because this is so small, it is set to false