Facemorph
Class Warp

java.lang.Object
  Facemorph.Warp

All Implemented Interfaces:

java.io.Serializable

Direct Known Subclasses:

LinearWarp, MultiscaleWarp, MultiscaleWarpRB, TPSWarp

public abstract class Warp
extends java.lang.Object
implements java.io.Serializable

Base warping class

See Also:

Serialized Form

Field Summary

int	height The height of the warping function
static int	LINEAR Linear warp constant
static int	MULTISCALE Multiscale (affine) warp constant
static int	MULTISCALERB Multiscale (rigid body)warp constant
static int	TPS Thin plate spline warp constant
int	width The width of the warping function
float[]	xshift The array of xshifts
float[]	yshift The array of y-shifts

Constructor Summary

Warp()
Construct an empty warp

Method Summary

void	concatenate(Warp warp) Concatenates warp with this
void	concatenate(Warp warp1, Warp warp2) Concatenates the two warps and puts the result in this
void	convert(FloatImage Wx, FloatImage Wy) Converts this warp into the two FloatImages
static Warp	createWarp(int type, int w, int h, int ow, int oh, boolean r) Factory method for creating a warp of the specified type
float[]	get(int x, int y) Gets the value of this warp at (x,y)
int	getHeight() Gets the height of this warp
int	getWidth() Gets the width of this warp
abstract boolean	interpolate(int n, java.awt.geom.Point2D.Float[] source, java.awt.geom.Point2D.Float[] dest, boolean borders) Method to interpolate the warping function
boolean	interpolate(int n, java.awt.geom.Point2D.Float[] source, java.awt.geom.Point2D.Float[] dest, boolean borders, boolean overlap) Interpolates the given 2D values across the 2D space, by performing setup then calling interpolate()
boolean	interpolate(Template tmp1, Template tmp2, boolean sample, boolean borders) Interpolate this warping function from tmp1 to tmp2
boolean	interpolate(Template tmp1, Template tmp2, boolean sample, boolean borders, boolean overlap) Interpolate this warping function from tmp1 to tmp2
int	iterate(int n, java.awt.geom.Point2D.Float[] source, java.awt.geom.Point2D.Float[] dest, Warp tmp, float minJ, boolean borders) Implements an overlap control algorithm
float	overlap(float minJ) Calculates the scaling parameter to ensure the Jacobean >= minJ and applies (if less than 1) it to this warp
void	reconvert(FloatImage Wx, FloatImage Wy) Converts the warp sepcified as two float images into this
float[]	sample(float x, float y) Samples this warp at non-intege pixels using bi-linear interpolation
static java.awt.Color	sample(int[] pix, int width, int height, float x, float y) Sample from a non-integer point in an image (as an int array) using linear interpolated sampling
void	scaleWarp(float scaleFactor) Scales a warp by the amount specified
void	set(int x, int y, float X, float Y) Sets the value of this warp at the specified point
int	stripPoints(java.awt.geom.Point2D.Float[] source, java.awt.geom.Point2D.Float[] dest, java.awt.geom.Point2D.Float[] newSource, java.awt.geom.Point2D.Float[] newDest, int n, int w1, int h1, int w2, int h2) Removes points from both lists that are outside the image borders of their respective images
FloatImage	warpFloatImage(FloatImage fimg) Warp a FloatImage using this warping function
java.awt.image.BufferedImage	warpImage(java.awt.image.BufferedImage img) Warp an image using this warping function
java.awt.Image	warpImage(java.awt.Image img, java.awt.image.ImageObserver ob) Warp an image using this warping function
int[]	warpImage(java.awt.Image img, int twidth, int theight) Warp an image using this warping function

Methods inherited from class java.lang.Object

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

Field Detail

xshift

public float[] xshift

The array of xshifts

yshift

public float[] yshift

The array of y-shifts

width

public int width

The width of the warping function

height

public int height

The height of the warping function

MULTISCALE

public static final int MULTISCALE

Multiscale (affine) warp constant

See Also:

Constant Field Values

LINEAR

public static final int LINEAR

Linear warp constant

See Also:

Constant Field Values

TPS

public static final int TPS

Thin plate spline warp constant

See Also:

Constant Field Values

MULTISCALERB

public static final int MULTISCALERB

Multiscale (rigid body)warp constant

See Also:

Constant Field Values

Constructor Detail

Warp

public Warp()

Construct an empty warp

Method Detail

createWarp

public static Warp createWarp(int type,
                              int w,
                              int h,
                              int ow,
                              int oh,
                              boolean r)

Factory method for creating a warp of the specified type

Parameters:

type - the type of warp to make, one of MULTISCALE, LINEAR, TPS or MULTISCALERB

w - the output width

h - the output height

ow - the input width

oh - the input height

r - indicates reverse of the triangulation (i.e. triangulate the output not the input)

Returns:

returns the warp of the requested type

interpolate

public boolean interpolate(int n,
                           java.awt.geom.Point2D.Float[] source,
                           java.awt.geom.Point2D.Float[] dest,
                           boolean borders,
                           boolean overlap)

Interpolates the given 2D values across the 2D space, by performing setup then calling interpolate()

Parameters:

n - The number of constraints to interpolate

source - The array of start points

dest - The array of end points

borders - Specifies whether to fix the points on the borders

overlap - indicates that overlap control should be used

Returns:

true of the interpolation was successful, false otherwise

interpolate

public boolean interpolate(Template tmp1,
                           Template tmp2,
                           boolean sample,
                           boolean borders,
                           boolean overlap)

Interpolate this warping function from tmp1 to tmp2

Parameters:

tmp1 - Source template

tmp2 - target template

sample - should the contour samples be used, or just the control points

borders - Should we pin the border down

overlap - indicates that overlap control should be used

Returns:

true if interpolation successful

getWidth

public int getWidth()

Gets the width of this warp

Returns:

The width

getHeight

public int getHeight()

Gets the height of this warp

Returns:

The height

get

public float[] get(int x,
                   int y)

Gets the value of this warp at (x,y)

Parameters:

x - The x-coord

y - The y-coord

Returns:

The value at (x,y)

set

public void set(int x,
                int y,
                float X,
                float Y)

Sets the value of this warp at the specified point

Parameters:

x - The x-coord

y - The y-coord

X - The X value

Y - The Y value

interpolate

public abstract boolean interpolate(int n,
                                    java.awt.geom.Point2D.Float[] source,
                                    java.awt.geom.Point2D.Float[] dest,
                                    boolean borders)

Method to interpolate the warping function

Parameters:

n - The number of constraints

source - The start points

dest - The target points

borders - if true fix the borders

Returns:

true if the interpolation was successful

interpolate

public boolean interpolate(Template tmp1,
                           Template tmp2,
                           boolean sample,
                           boolean borders)

Interpolate this warping function from tmp1 to tmp2

Parameters:

tmp1 - Source template

tmp2 - target template

sample - should the contour samples be used, or just the control points

borders - Should we pin the border down

Returns:

true if interpolation successful

stripPoints

public int stripPoints(java.awt.geom.Point2D.Float[] source,
                       java.awt.geom.Point2D.Float[] dest,
                       java.awt.geom.Point2D.Float[] newSource,
                       java.awt.geom.Point2D.Float[] newDest,
                       int n,
                       int w1,
                       int h1,
                       int w2,
                       int h2)

Removes points from both lists that are outside the image borders of their respective images

Parameters:

source - List of source points

dest - List of destination points

newSource - output source point list

newDest - output de3stination point list

n - the number of input points

w1 - he width of image 1

h1 - The height of image 1

w2 - The width of image 2

h2 - The height of image 2

Returns:

The number of points in the stripped set

sample

public static java.awt.Color sample(int[] pix,
                                    int width,
                                    int height,
                                    float x,
                                    float y)

Sample from a non-integer point in an image (as an int array) using linear interpolated sampling

Parameters:

pix - An array of the images pixels as RGB ints

width - The width of the image

height - The height of theimage

x - The x-coord of the sample point

y - Te y-coord of the sample point

Returns:

The sampled colour

sample

public float[] sample(float x,
                      float y)

Samples this warp at non-intege pixels using bi-linear interpolation

Parameters:

x - the x-coord to sample

y - the y-coord to sample

Returns:

the target of the warping function as an (x,y) pair

warpImage

public int[] warpImage(java.awt.Image img,
                       int twidth,
                       int theight)

Warp an image using this warping function

Parameters:

img - The image to warp

twidth - The width of the input image

theight - The height of the input image

Returns:

The warped image as an array of RGB int values

warpImage

public java.awt.Image warpImage(java.awt.Image img,
                                java.awt.image.ImageObserver ob)

Warp an image using this warping function

Parameters:

img - The image to warp

ob - An image observer object

Returns:

The warped image

warpImage

public java.awt.image.BufferedImage warpImage(java.awt.image.BufferedImage img)

Warp an image using this warping function

Parameters:

img - The image to warp

Returns:

The warped image

warpFloatImage

public FloatImage warpFloatImage(FloatImage fimg)

Warp a FloatImage using this warping function

Parameters:

fimg - The image to warp

Returns:

The warped image

concatenate

public void concatenate(Warp warp)

Concatenates warp with this

Parameters:

warp - the warp to concatenate

concatenate

public void concatenate(Warp warp1,
                        Warp warp2)

Concatenates the two warps and puts the result in this

Parameters:

warp1 - the first warp

warp2 - the second warp

convert

public void convert(FloatImage Wx,
                    FloatImage Wy)

Converts this warp into the two FloatImages

Parameters:

Wx - the output x shifts

Wy - the output y shifts

reconvert

public void reconvert(FloatImage Wx,
                      FloatImage Wy)

Converts the warp sepcified as two float images into this

Parameters:

Wx - warp x-component

Wy - warp y component

iterate

public int iterate(int n,
                   java.awt.geom.Point2D.Float[] source,
                   java.awt.geom.Point2D.Float[] dest,
                   Warp tmp,
                   float minJ,
                   boolean borders)

Implements an overlap control algorithm

Parameters:

n - number of points

source - source points

dest - target points

tmp - tempory warp workspace

minJ - minimum allowed value of the Jacobean in any partial warp

borders - indicates if borders should be pinned down

Returns:

returns 1

overlap

public float overlap(float minJ)

Calculates the scaling parameter to ensure the Jacobean >= minJ and applies (if less than 1) it to this warp

Parameters:

minJ - the minimum allowed value of the Jacobean

Returns:

returns the scaling value

scaleWarp

public void scaleWarp(float scaleFactor)

Scales a warp by the amount specified

Parameters:

scaleFactor - the scale factor to apply to this warp