Facemorph.psm
Class PSM

java.lang.Object
  Facemorph.psm.PSM

public class PSM
extends java.lang.Object

PSM implementation using steerable pyramids (doesn't work very well, the PSMNode class is preferred)

See Also:

PSMNode

Constructor Summary

PSM()
PSM constructor

PSM(int start, BigMat mst, DiagonalGMM[] appGauss, DiagonalGaussian[] gauss, boolean[] visited, java.util.ArrayList<Filter> filters, Filter lowpass, int w, int h, int levs)
PSM constructor

Method Summary

void	batchTest(java.lang.String outfold, java.util.ArrayList<java.lang.String> imageNames, java.util.ArrayList<java.lang.String> templateNames, int w)
double	calculateFeatureProbability(SteerablePyramid pyr, float x, float y) Calculate the probability of this feature
FloatImage	evaluateProbability(FloatImage source) Evaluate the probability using Steerable pyramid
FloatImage	evaluateProbability(SteerablePyramid pyr) Evaluate the probability by recursing down and up the tree
static int	findBestRoot(BigMat tree, BigMat graph) Find the best root of the tree
java.util.ArrayList<java.awt.geom.Point2D.Float>	getFeatures() Get the feature points
java.util.ArrayList<java.awt.geom.Point2D.Float>	getFeatures(FloatImage source) Get the feature points in the image
void	getFeatures(int x, int y, java.util.ArrayList<java.awt.geom.Point2D.Float> points) Get the features
static double[]	getFeatureVector(SteerablePyramid pyr, float x, float y) Get the feature vector from a steerable pyramid
static void	main(java.lang.String[] args)
static BigMat	minimumSpanningTree(BigMat graph) Calculate the MST
static double	ncc(double[] v1, double[] v2) The normalised cross correlation
void	read(java.io.InputStream in) Read from an input stream
void	read(java.io.StreamTokenizer st) Read from a StreamTokenizer
void	read(java.lang.String fileName) Read from a file
static PSM	train(java.util.ArrayList<java.lang.String> imageNames, java.util.ArrayList<java.lang.String> templateNames, int[] pointList, int w, java.util.ArrayList<Filter> filters, Filter lowpass, int maxLev) PSM training method
void	write(java.io.PrintStream ps) Write to a PrintStream
void	write(java.lang.String fileName) Write PSM to a file

Methods inherited from class java.lang.Object

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

Constructor Detail

PSM

public PSM()

PSM constructor

PSM

public PSM(int start,
           BigMat mst,
           DiagonalGMM[] appGauss,
           DiagonalGaussian[] gauss,
           boolean[] visited,
           java.util.ArrayList<Filter> filters,
           Filter lowpass,
           int w,
           int h,
           int levs)

PSM constructor

Parameters:

start - the start node

mst - the minimum spanning tree

appGauss - the appearance models for the features

gauss - the point distributions

visited - indicates which nodes have been visited

filters - the filters to use to build the pyramids

lowpass - the lowpass filters to use to build the pyramids

w - the width

h - the height

levs - the levels

Method Detail

write

public void write(java.lang.String fileName)

Write PSM to a file

Parameters:

fileName - the file to write to

write

public void write(java.io.PrintStream ps)

Write to a PrintStream

Parameters:

ps - the PrintStream to write to

read

public void read(java.lang.String fileName)

Read from a file

Parameters:

fileName - the file to read from

read

public void read(java.io.InputStream in)
          throws java.io.IOException

Read from an input stream

Parameters:

in - the stream to read from

Throws:

java.io.IOException

read

public void read(java.io.StreamTokenizer st)
          throws java.io.IOException

Read from a StreamTokenizer

Parameters:

st - the StreamTokenizer to read from

Throws:

java.io.IOException

train

public static PSM train(java.util.ArrayList<java.lang.String> imageNames,
                        java.util.ArrayList<java.lang.String> templateNames,
                        int[] pointList,
                        int w,
                        java.util.ArrayList<Filter> filters,
                        Filter lowpass,
                        int maxLev)

PSM training method

Parameters:

imageNames - the names of the training images

templateNames - the names of the training templates

pointList - the list of points to use from each template

w - the training width

filters - the filters to use to build the pyramid

lowpass - the low pass filter to use to build the pyramid

maxLev - the maximum level to build the pyramid to

Returns:

return the trained PSM

findBestRoot

public static int findBestRoot(BigMat tree,
                               BigMat graph)

Find the best root of the tree

Parameters:

tree - the mst

graph - the original completed weighted edge graph

Returns:

returns the best root (in some sense)

minimumSpanningTree

public static BigMat minimumSpanningTree(BigMat graph)

Calculate the MST

Parameters:

graph - the Graph to find the MST for

Returns:

returns the MST

evaluateProbability

public FloatImage evaluateProbability(FloatImage source)

Evaluate the probability using Steerable pyramid

Parameters:

source - the source image

Returns:

returns the probability image

getFeatureVector

public static double[] getFeatureVector(SteerablePyramid pyr,
                                        float x,
                                        float y)

Get the feature vector from a steerable pyramid

Parameters:

pyr - the pyramid

x - the x-coors

y - the y-coord

Returns:

return the feature vector from the pyramid

calculateFeatureProbability

public double calculateFeatureProbability(SteerablePyramid pyr,
                                          float x,
                                          float y)

Calculate the probability of this feature

Parameters:

pyr - the steerable pyramid

x - x-coord

y - y-coord

Returns:

the probability of the feature at (x,y)

evaluateProbability

public FloatImage evaluateProbability(SteerablePyramid pyr)

Evaluate the probability by recursing down and up the tree

Parameters:

pyr - the pyramid to use

Returns:

returns the probabilities as an image

getFeatures

public void getFeatures(int x,
                        int y,
                        java.util.ArrayList<java.awt.geom.Point2D.Float> points)

Get the features

Parameters:

x - the x-coord

y - the y-coord

points - the point list to add the features to

getFeatures

public java.util.ArrayList<java.awt.geom.Point2D.Float> getFeatures(FloatImage source)

Get the feature points in the image

Parameters:

source - the input image

Returns:

the estimated feature locations

getFeatures

public java.util.ArrayList<java.awt.geom.Point2D.Float> getFeatures()

Get the feature points

Returns:

return the most likely feature points

ncc

public static double ncc(double[] v1,
                         double[] v2)

The normalised cross correlation

Parameters:

v1 - the first vector

v2 - the second vector

Returns:

returns the ncc between the two vectors

batchTest

public void batchTest(java.lang.String outfold,
                      java.util.ArrayList<java.lang.String> imageNames,
                      java.util.ArrayList<java.lang.String> templateNames,
                      int w)

Parameters:

outfold -

imageNames -

templateNames -

w -

main

public static void main(java.lang.String[] args)

Parameters:

args -