org.openimaj.image.model

Class EigenImages

java.lang.Object

org.openimaj.image.model.EigenImages

All Implemented Interfaces:

FeatureExtractor<DoubleFV,FImage>, ReadableBinary, ReadWriteableBinary, WriteableBinary, BatchTrainer<FImage>

@Reference(type=Inproceedings,
           author={"Turk, M.A.","Pentland, A.P."},
           title="Face recognition using eigenfaces",
           year="1991",
           booktitle="Computer Vision and Pattern Recognition, 1991. Proceedings CVPR \'91., IEEE Computer Society Conference on",
           pages={"586 ","591"},
           month="jun",
           number="",
           volume="",
           customData={"keywords","eigenfaces;eigenvectors;face images;face recognition system;face space;feature space;human faces;two-dimensional recognition;unsupervised learning;computerised pattern recognition;eigenvalues and eigenfunctions;","doi","10.1109/CVPR.1991.139758"})
public class EigenImages
extends Object
implements BatchTrainer<FImage>, FeatureExtractor<DoubleFV,FImage>, ReadWriteableBinary

Implementation of EigenImages. Can be used for things like face recognition ala the classic EigenFaces algorithm.

Fundamentally, the EigenImages technique is a way to perform dimensionality reduction on an image using PCA. This implementation can be trained through the BatchTrainer interface (which will internally normalise the data and perform PCA). Once trained, instances can be used as FeatureExtractors to extract low(er) dimensional features from images.

Methods are also provided to reconstruct an image from its feature vector (see and #reconstruct(double[]), and to visualise a specific principal component as an image (see visualisePC(int).

Author:

Jonathon Hare (jsh2@ecs.soton.ac.uk)

Constructor Summary

Constructors

Modifier	Constructor and Description
protected	EigenImages() For serialisation
	EigenImages(int numComponents) Construct with the given number of principal components.

Method Summary

Methods

Modifier and Type	Method and Description
byte[]	binaryHeader() Header for binary input.
DoubleFV	extractFeature(FImage img) Extract features from an object and return them.
int	getNumComponents() Get the number of PCA components selected by this EigenImages object.
void	readBinary(DataInput in) Read internal state from in.
FImage	reconstruct(double[] weights) Reconstruct an image from a weight vector
FImage	reconstruct(DoubleFV weights) Reconstruct an image from a weight vector
String	toString()
void	train(List<? extends FImage> data) Train the object with the given data.
FImage	visualisePC(int pc) Draw a principal component as an image.
void	writeBinary(DataOutput out) Write the content of this as binary to out.

Methods inherited from class java.lang.Object

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

Constructor Detail

EigenImages

protected EigenImages()

For serialisation

EigenImages

public EigenImages(int numComponents)

Construct with the given number of principal components.

Parameters:

numComponents - the number of PCs

Method Detail

extractFeature

public DoubleFV extractFeature(FImage img)

Description copied from interface: FeatureExtractor

Extract features from an object and return them.

Specified by:

extractFeature in interface FeatureExtractor<DoubleFV,FImage>

Parameters:

img - the object to extract from

Returns:

the extracted feature

train

public void train(List<? extends FImage> data)

Description copied from interface: BatchTrainer

Train the object with the given data. BatchTrainers should reset the underlying model everytime this method is called.

Specified by:

train in interface BatchTrainer<FImage>

Parameters:

data - the training data

reconstruct

public FImage reconstruct(DoubleFV weights)

Reconstruct an image from a weight vector

Parameters:

weights - the weight vector

Returns:

the reconstructed image

reconstruct

public FImage reconstruct(double[] weights)

Reconstruct an image from a weight vector

Parameters:

weights - the weight vector

Returns:

the reconstructed image

visualisePC

public FImage visualisePC(int pc)

Draw a principal component as an image. The image will be normalised so it can be displayed correctly.

Parameters:

pc - the index of the PC to draw.

Returns:

an image showing the PC.

readBinary

public void readBinary(DataInput in)
                throws IOException

Description copied from interface: ReadableBinary

Read internal state from in.

Specified by:

readBinary in interface ReadableBinary

Parameters:

in - source to read from.

Throws:

IOException - an error reading input

binaryHeader

public byte[] binaryHeader()

Description copied from interface: ReadableBinary

Header for binary input. Will be automatically read by IOUtils when using readBinary().

Specified by:

binaryHeader in interface ReadableBinary

Specified by:

binaryHeader in interface WriteableBinary

Returns:

header

writeBinary

public void writeBinary(DataOutput out)
                 throws IOException

Description copied from interface: WriteableBinary

Write the content of this as binary to out.

Specified by:

writeBinary in interface WriteableBinary

Parameters:

out - sink to write to

Throws:

IOException - an error writing to out

toString

public String toString()

Overrides:

toString in class Object

getNumComponents

public int getNumComponents()

Get the number of PCA components selected by this EigenImages object.

Returns:

the number of PCA components.