cbr.ccv
Class CCVExtractor

java.lang.Object
  cbr.ccv.CCVExtractor

All Implemented Interfaces:

FeatureExtractor

public class CCVExtractor

extends Object

implements FeatureExtractor

This class extracts color coherence vectors from images.

Version:

"$Id: CCVExtractor.java 1913 2007-08-08 02:41:53Z jpeters $"

Author:

Volker Roth

Field Summary

protected ArrayList	activeareas_ The array of active areas.
protected ArrayList	activespans_ The array of active spans.
protected ArrayList	areas_ The list of areas whose analysis is in progress.
protected int[]	coherent_ The per color counts of coherent pixels.
protected ArrayList	freeareas_ The array of unused areas.
protected ArrayList	freespans_ The array of unused spans.
protected int[]	incoherent_ The per color counts of incoherent pixels.
protected ArrayList	spans_ The list of spans whose analysis is in progress.
protected float	tau_ The tau value in percent of total pixel count.

Constructor Summary

CCVExtractor()
Creates a fresh instance.

CCVExtractor(float tau)
Creates a fresh instance that uses the given tau to distinguish between coherent and incoherent pixel areas.

Method Summary

FeatureVector	extract(byte[] px, int w, int h) This method extracts the CCV from a byte array of RGB222 pixels.
FeatureVector	extract(Image im) Extracts the color coherence vector from the given image.
FeatureVector	extract(int[] px, int width, int height) Extracts the color coherence vector from the given array of RGB888 pixels.
protected void	freeArea(Area area)
protected Area	newArea(int color, int size)
protected void	scan(List list, byte[] px, int off, int width) Fills the given list instance with the list of spans initialised from the given pixel buffer starting at the given offset.

Methods inherited from class java.lang.Object

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

Field Detail

freespans_

protected ArrayList freespans_

The array of unused spans. Spans are re-used for better performance. This avoids expensive creation of objects.

freeareas_

protected ArrayList freeareas_

The array of unused areas. Areas are re-used for better performance. This avoids expensive creation of objects.

activespans_

protected ArrayList activespans_

The array of active spans.

activeareas_

protected ArrayList activeareas_

The array of active areas.

spans_

protected ArrayList spans_

The list of spans whose analysis is in progress.

areas_

protected ArrayList areas_

The list of areas whose analysis is in progress.

coherent_

protected int[] coherent_

The per color counts of coherent pixels.

incoherent_

protected int[] incoherent_

The per color counts of incoherent pixels.

tau_

protected float tau_

The tau value in percent of total pixel count. The default value is 1% (0.01).

Constructor Detail

CCVExtractor

public CCVExtractor()

Creates a fresh instance.

CCVExtractor

public CCVExtractor(float tau)

Creates a fresh instance that uses the given tau to distinguish between coherent and incoherent pixel areas. The tau value must be in the range [0,1] and stands for the minimum percentage of the overall image size in pixels that a continuous pixel area must represent in order to be classified as coherent. A good value of tau is 0,01 (1% of the image area). This is also the defaut value in case the default constructor is called.

Parameters:

tau - The tau value.

Method Detail

extract

public FeatureVector extract(Image im)
                      throws InterruptedException,
                             ImageException

Extracts the color coherence vector from the given image. The image is blurred and quantised before the vector is extracted.

Specified by:

extract in interface FeatureExtractor

Parameters:

im - The image from which the CCV is extacted.

Returns:

The color coherence vector of the given image.

Throws:

InterruptedException - if an interrupt occurs while the image's pixel data is loaded.

ImageException - if for instance the dimensions of the given image cannot be determined.

extract

public FeatureVector extract(int[] px,
                             int width,
                             int height)

Extracts the color coherence vector from the given array of RGB888 pixels. The pixels are blurred and quantised before extraction.

Specified by:

extract in interface FeatureExtractor

Parameters:

px - The byte array of RGB222 pixel values of the image from which the CCV is extracted.

Returns:

The color coherence vector of the given image.

extract

public FeatureVector extract(byte[] px,
                             int w,
                             int h)

This method extracts the CCV from a byte array of RGB222 pixels. The pixel array should already be blurred. This method is primarily called internally. It will be rarely used by an external instance but for testing purposes.

The caller has to make sure that the pixel values are in the range [0,63]. Any pixel outside this range will cause an ArrayIndexOutOfBoundsException to be thrown in the course of this method call. The easiest way to assure this range is to use the Quantizer.

Specified by:

extract in interface FeatureExtractor

Parameters:

px - The byte array of RGB222 pixel values of the image from which the CCV is extracted.

w - The width of the image in pixels.

h - The height of the image in pixels.

Returns:

The color coherence vector of the given image.

scan

protected void scan(List list,
                    byte[] px,
                    int off,
                    int width)

Fills the given list instance with the list of spans initialised from the given pixel buffer starting at the given offset. Exactly width pixels are consumed.

newArea

protected Area newArea(int color,
                       int size)

freeArea

protected void freeArea(Area area)