net.sf.ij_plugins.im3d.grow

Class SRG

java.lang.Object

net.sf.ij_plugins.util.progress.DefaultProgressReporter

net.sf.ij_plugins.im3d.grow.SRG

All Implemented Interfaces:

ProgressReporter

public final class SRG
extends DefaultProgressReporter

Seeded region growing algorithm based on article by Rolf Adams and Leanne Bischof, "Seeded Region Growing", IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 16, no. 6, June 1994.

The algorithm assumes that seeds for objects and the background be provided. Seeds are used to compute initial mean gray level for each region. The condition of growth is difference of a gray level of a candidate pixel and mean grey level intensity of a neighboring region. At each step of the algorithm a candidate with a smallest difference to some neighboring region is added to that region and all neighboring points of that that are not yet assigned to any region are added to candidate list.

Part of the image that will be segmented can be restricted by setting a mask.

Progress of segmentation can be recorded on a animation stack. To enable recording set numberOfAnimationFrames to a value larger than 0.

An example of segmenting an image is below. Seeds are set for three regions: background, blob 1, and blob 2.

 // Setup growing
 final ByteProcessor image = ...;
 final Point[][] seeds = {
         {new Point(107, 144)}, // Background
         {new Point(91, 159)},  // Blob 1
         {new Point(119, 143)}, // Blob 2
     };
 final SRG srg = new SRG();
 srg.setImage(image);
 srg.setSeeds(seeds);
 srg.setNumberOfAnimationFrames(50);
 // Run growing
 srg.run();
 // Extract results
 final ByteProcessor regionMask = srg.getRegionMarkers();
 final ImageStack animationStack = srg.getAnimationStack();
 

Author:

Jarek Sacha

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
protected static class	net.sf.ij_plugins.im3d.grow.SRG2DBase.RegionInfo

Field Summary

Fields

Modifier and Type	Field and Description
protected int	xSize
protected int	ySize

Constructor Summary

Constructors

Constructor and Description
SRG()

Method Summary

Methods

Modifier and Type	Method and Description
protected double	distanceFromMean(int offset, net.sf.ij_plugins.im3d.grow.SRG2DBase.RegionInfo regionInfo)
ij.ImageStack	getAnimationStack() Return progress of growing.
ij.process.ByteProcessor	getRegionMarkers()
protected void	initializeImageStructures()
protected void	initializeStructures()
protected net.sf.ij_plugins.im3d.grow.SRG2DBase.RegionInfo	newRegionInfo(int originalSeedID)
void	run() Perform region growing.
void	setImage(ij.process.ByteProcessor image) Set image to be segmented.
void	setImage(ij.process.FloatProcessor image) Set image to be segmented.
void	setImage(ij.process.ImageProcessor image) Set image to be segmented.
void	setImage(ij.process.ShortProcessor image) Set image to be segmented.
void	setMask(ij.process.ByteProcessor mask) Set ROI mask, if not set full image is processed.
void	setNumberOfAnimationFrames(int numberOfAnimationFrames) Sets how many animation frames should be recorded.
void	setSeeds(ij.process.ByteProcessor seeds) Set region seeds.
static ij.process.ByteProcessor	toSeedImage(java.awt.Point[][] seeds, int width, int height) Convert array of point seeds to a seed image.

Methods inherited from class net.sf.ij_plugins.util.progress.DefaultProgressReporter

addProgressListener, currentProgress, notifyProgressListeners, notifyProgressListeners, notifyProgressListeners, removeAllProgressListener, removeProgressListener, setCurrentProgress

Methods inherited from class java.lang.Object

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

Field Detail

xSize

protected int xSize

ySize

protected int ySize

Constructor Detail

SRG

public SRG()

Method Detail

setImage

public void setImage(ij.process.ImageProcessor image)

Set image to be segmented. Supported types are: ByteProcessor, ShortProcessor, and FloatProcessor.

Parameters:

image - image.

setImage

public void setImage(ij.process.ByteProcessor image)

Set image to be segmented.

Parameters:

image - image.

setImage

public void setImage(ij.process.ShortProcessor image)

Set image to be segmented.

Parameters:

image - image.

setImage

public void setImage(ij.process.FloatProcessor image)

Set image to be segmented.

Parameters:

image - image.

distanceFromMean

protected double distanceFromMean(int offset,
                      net.sf.ij_plugins.im3d.grow.SRG2DBase.RegionInfo regionInfo)

initializeImageStructures

protected void initializeImageStructures()

newRegionInfo

protected net.sf.ij_plugins.im3d.grow.SRG2DBase.RegionInfo newRegionInfo(int originalSeedID)

setSeeds

public void setSeeds(ij.process.ByteProcessor seeds)

Set region seeds. The first index in the seeds array refers to the region. Second index is a seeds with a region.

Parameters:

seeds - region seeds.

See Also:

toSeedImage(java.awt.Point[][], int, int)

setMask

public void setMask(ij.process.ByteProcessor mask)

Set ROI mask, if not set full image is processed.

Parameters:

mask - Mask of the region of interest used for processing.

setNumberOfAnimationFrames

public void setNumberOfAnimationFrames(int numberOfAnimationFrames)

Sets how many animation frames should be recorded. If less then 1 animation will not be recorded.

Parameters:

numberOfAnimationFrames - number of animation frames.

getRegionMarkers

public ij.process.ByteProcessor getRegionMarkers()

getAnimationStack

public ij.ImageStack getAnimationStack()

Return progress of growing.

Returns:

animation stack with region markers.

toSeedImage

public static ij.process.ByteProcessor toSeedImage(java.awt.Point[][] seeds,
                                   int width,
                                   int height)

Convert array of point seeds to a seed image. The first index in the seeds array refers to the region. Region number if the first index + 1. Second index is a seeds with a region.

Conditions: .

1. Argument seeds cannot be link null.
2. There must be at least 2 regions.
3. There must be at most 253 regions.
4. Each region has to have at least one seed point.
5. All seed points have to be unique.

Parameters:

seeds - region seeds.

width - width of the seed image

height - height of the seed image

Returns:

seed image.

run

public void run()

Perform region growing.

initializeStructures

protected void initializeStructures()