org.openimaj.image.processing.face.tracking.clm

Class CLMFaceTracker

java.lang.Object

org.openimaj.image.processing.face.tracking.clm.CLMFaceTracker

public class CLMFaceTracker
extends Object

CLM-based face tracker

Author:

David Dupplaw (dpd@ecs.soton.ac.uk)

Field Summary

Fields

Modifier and Type	Field and Description
double	clamp Number of standard deviations from the mean face to allow in the model
int[][]	connections The face connections
boolean	fcheck Whether to use the face check (using pixels as a face classifier)
int	fpd Number of frames on which to force a redetection
double	fTol Model fitting optimisation tolerance
MultiTracker	model The tracker to use
int	nIter Number of iterations to use for model fitting
float	scale The scale at which to process the video
float	searchAreaSize The size of the search area for redetection (template matching)
int[][]	triangles The face mesh
int[]	wSize1 Search window size while tracking
int[]	wSize2 Search window size when initialising after a failed track

Constructor Summary

Constructors

Constructor and Description
CLMFaceTracker() Instantiates a tracker for tracking faces based on some default values and models.

Method Summary

Modifier and Type	Method and Description
static void	drawFaceModel(MBFImage image, MultiTracker.TrackedFace f, boolean drawTriangles, boolean drawConnections, boolean drawPoints, boolean drawSearchArea, boolean drawBounds, int[][] triangles, int[][] connections, float scale, Float[] boundingBoxColour, Float[] meshColour, Float[] connectionColour, Float[] pointColour) Draw onto the given image, the given face model.
void	drawModel(MBFImage image, boolean drawTriangles, boolean drawConnections, boolean drawPoints, boolean drawSearchArea, boolean drawBounds) Draw the model onto the image
Float[]	getBoundingBoxColour()
Float[]	getConnectionColour()
MultiTracker.TrackerVars	getInitialVars() Returns the initial variables that will be used by the tracker for each found face.
Float[]	getMeshColour()
MultiTracker	getModelTracker() Returns the model tracker
Float[]	getPointColour()
int[][]	getReferenceConnections() Get the reference connections
int[][]	getReferenceTriangles() Get the reference triangles
Float[]	getSearchAreaColour()
float	getSearchAreaSize()
List<MultiTracker.TrackedFace>	getTrackedFaces()
static List<Triangle>	getTriangles(Jama.Matrix shape, Jama.Matrix visi, int[][] triangles) Get the triangle mesh corresponding to a tracked face.
List<Triangle>	getTriangles(MultiTracker.TrackedFace face) Get the triangle mesh corresponding to a tracked face.
void	initialiseFaceModel(MultiTracker.TrackedFace face) Initialises the face model for the tracked face by calling MultiTracker.initShape(Rectangle, Matrix, Matrix) with the rectangle of MultiTracker.TrackedFace.redetectedBounds and the face shape and the reference shape.
void	reset() Force a reset on the next frame to be tracked.
void	setBoundingBoxColour(Float[] boundingBoxColour)
void	setConnectionColour(Float[] connectionColour)
void	setMeshColour(Float[] meshColour)
void	setPointColour(Float[] pointColour)
void	setRedetectEvery(int nFrames) Set the number of frames after which a redection is forced by the tracker.
void	setSearchAreaColour(Float[] searchAreaColour)
void	setSearchAreaSize(float searchAreaSize)
void	track(FImage im) Track the face in the given frame.
void	track(MBFImage frame) Track the face in the given frame.

Methods inherited from class java.lang.Object

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

Field Detail

model

public MultiTracker model

The tracker to use

triangles

public int[][] triangles

The face mesh

connections

public int[][] connections

The face connections

scale

public float scale

The scale at which to process the video

fcheck

public boolean fcheck

Whether to use the face check (using pixels as a face classifier)

fpd

public int fpd

Number of frames on which to force a redetection

wSize1

public int[] wSize1

Search window size while tracking

wSize2

public int[] wSize2

Search window size when initialising after a failed track

nIter

public int nIter

Number of iterations to use for model fitting

clamp

public double clamp

Number of standard deviations from the mean face to allow in the model

fTol

public double fTol

Model fitting optimisation tolerance

searchAreaSize

public float searchAreaSize

The size of the search area for redetection (template matching)

Constructor Detail

CLMFaceTracker

public CLMFaceTracker()

Instantiates a tracker for tracking faces based on some default values and models.

Method Detail

track

public void track(MBFImage frame)

Track the face in the given frame.

Parameters:

frame - The frame

track

public void track(FImage im)

Track the face in the given frame.

Parameters:

im - The frame

reset

public void reset()

Force a reset on the next frame to be tracked.

drawModel

public void drawModel(MBFImage image,
                      boolean drawTriangles,
                      boolean drawConnections,
                      boolean drawPoints,
                      boolean drawSearchArea,
                      boolean drawBounds)

Draw the model onto the image

Parameters:

image - The image to draw onto

drawTriangles - Whether to draw the triangles

drawConnections - Whether to draw the connections

drawPoints - Whether to draw the points

drawSearchArea - Whether to draw the search area

drawBounds - Whether to draw the bounds

drawFaceModel

public static void drawFaceModel(MBFImage image,
                                 MultiTracker.TrackedFace f,
                                 boolean drawTriangles,
                                 boolean drawConnections,
                                 boolean drawPoints,
                                 boolean drawSearchArea,
                                 boolean drawBounds,
                                 int[][] triangles,
                                 int[][] connections,
                                 float scale,
                                 Float[] boundingBoxColour,
                                 Float[] meshColour,
                                 Float[] connectionColour,
                                 Float[] pointColour)

Draw onto the given image, the given face model.

Parameters:

image - The image to draw onto

f - The face model to draw

drawTriangles - Whether to draw the triangles

drawConnections - Whether to draw the connections

drawPoints - Whether to draw the points

drawSearchArea - Whether to draw the search area

drawBounds - Whether to draw the bounds

triangles - The reference triangles

connections - The reference connections

scale - The scale at which to draw

boundingBoxColour - Colour to draw the bounding box

meshColour - Colour to draw the mesh

connectionColour - Colour to draw the connections

pointColour - Colour to draw the points

getReferenceTriangles

public int[][] getReferenceTriangles()

Get the reference triangles

Returns:

The triangles

getReferenceConnections

public int[][] getReferenceConnections()

Get the reference connections

Returns:

The connections

getModelTracker

public MultiTracker getModelTracker()

Returns the model tracker

Returns:

The model tracker

getInitialVars

public MultiTracker.TrackerVars getInitialVars()

Returns the initial variables that will be used by the tracker for each found face.

Returns:

The initial tracker variables.

initialiseFaceModel

public void initialiseFaceModel(MultiTracker.TrackedFace face)

Initialises the face model for the tracked face by calling MultiTracker.initShape(Rectangle, Matrix, Matrix) with the rectangle of MultiTracker.TrackedFace.redetectedBounds and the face shape and the reference shape. Assumes that the bounds have been already set up.

Parameters:

face - The face to initialise

getSearchAreaSize

public float getSearchAreaSize()

Returns:

the searchAreaSize

setSearchAreaSize

public void setSearchAreaSize(float searchAreaSize)

Parameters:

searchAreaSize - the searchAreaSize to set

getConnectionColour

public Float[] getConnectionColour()

Returns:

the connectionColour

setConnectionColour

public void setConnectionColour(Float[] connectionColour)

Parameters:

connectionColour - the connectionColour to set

getPointColour

public Float[] getPointColour()

Returns:

the pointColour

setPointColour

public void setPointColour(Float[] pointColour)

Parameters:

pointColour - the pointColour to set

getMeshColour

public Float[] getMeshColour()

Returns:

the meshColour

setMeshColour

public void setMeshColour(Float[] meshColour)

Parameters:

meshColour - the meshColour to set

getBoundingBoxColour

public Float[] getBoundingBoxColour()

Returns:

the boundingBoxColour

setBoundingBoxColour

public void setBoundingBoxColour(Float[] boundingBoxColour)

Parameters:

boundingBoxColour - the boundingBoxColour to set

getSearchAreaColour

public Float[] getSearchAreaColour()

Returns:

the searchAreaColour

setSearchAreaColour

public void setSearchAreaColour(Float[] searchAreaColour)

Parameters:

searchAreaColour - the searchAreaColour to set

getTrackedFaces

public List<MultiTracker.TrackedFace> getTrackedFaces()

Returns:

the list of tracked faces from the previous call to track(MBFImage) or track(FImage).

getTriangles

public List<Triangle> getTriangles(MultiTracker.TrackedFace face)

Get the triangle mesh corresponding to a tracked face.

Parameters:

face - the MultiTracker.TrackedFace

Returns:

the mesh

getTriangles

public static List<Triangle> getTriangles(Jama.Matrix shape,
                                          Jama.Matrix visi,
                                          int[][] triangles)

Get the triangle mesh corresponding to a tracked face.

Parameters:

shape - the shape matrix

visi - the visibility matrix

triangles - the triangle definitions

Returns:

the mesh

setRedetectEvery

public void setRedetectEvery(int nFrames)

Set the number of frames after which a redection is forced by the tracker. Set it to -1 to avoid forcing any redetection.

Parameters:

nFrames - The number of frames.