net.imglib2.algorithm.gauss

Class AbstractGauss<T extends net.imglib2.type.numeric.NumericType<T>>

java.lang.Object

net.imglib2.algorithm.gauss.AbstractGauss<T>

Type Parameters:

T - - Defines the Type in which the actual computation is performed

Direct Known Subclasses:

GaussDouble, GaussFloat, GaussGeneral, GaussNativeType

public abstract class AbstractGauss<T extends net.imglib2.type.numeric.NumericType<T>>
extends Object

Author:

Stephan Preibisch

Constructor Summary

Constructors

Constructor and Description
AbstractGauss(double[] sigma, net.imglib2.RandomAccessible<T> input, net.imglib2.Interval inputInterval, net.imglib2.RandomAccessible<T> output, net.imglib2.Localizable outputOffset, net.imglib2.img.ImgFactory<T> factory, T type)

Method Summary

Modifier and Type	Method and Description
void	call()
protected void	computeKernel() Compute the Gaussian Kernel for all dimensions
protected SamplingLineIterator<T>	createInputLineSampler(int dim, net.imglib2.Interval range) An Iterator that samples a one dimensional line of input data for the current dimension.
protected WritableLineIterator<T>	createOutputLineWriter(int dim, net.imglib2.Interval range, SamplingLineIterator<T> inputLineSampler) An Iterator that samples a one dimensional line of output data for the current dimension.
net.imglib2.img.ImgFactory<T>	getFactory()
net.imglib2.RandomAccessible<T>	getInput()
net.imglib2.Interval	getInputInterval()
double[][]	getKernel()
int	getNumThreads()
net.imglib2.RandomAccessible<T>	getOutput()
net.imglib2.Localizable	getOutputOffset()
protected abstract net.imglib2.img.Img<T>	getProcessingLine(long size)
protected T	getProcessingType()
protected net.imglib2.Interval	getRange(int dim) The area for the output/temp that needs to be convolved, always relative to the input of the next convolution operation.
net.imglib2.RandomAccessible<T>	getResult()
double[]	getSigma()
protected net.imglib2.Interval	getTemporaryImgSize()
net.imglib2.img.Img<T>	getTmp1()
net.imglib2.img.Img<T>	getTmp2()
int	numDimensions()
protected void	processLine(SamplingLineIterator<T> input, double[] kernel) Compute the current line.
void	setNumThreads(int numThreads)
T	type()
protected void	updateInputLineSampler(SamplingLineIterator<T> a, net.imglib2.Interval range, long[] offset, net.imglib2.Localizable originalLocation) Updates the current SamplingLineIterator to the location of the new line that is processed and clears the processing line (set all values to zero)
protected void	updateOutputLineWriter(WritableLineIterator<T> a, net.imglib2.Interval range, long[] offset, net.imglib2.Localizable originalLocation) Updates the current WritableLineIterator to the location of the new line that is processed.
protected void	writeLine(WritableLineIterator<T> a, SamplingLineIterator<T> inputLineSampler) Writes the computed line back into the output/temp image.

Methods inherited from class java.lang.Object

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

Constructor Detail

AbstractGauss

public AbstractGauss(double[] sigma,
                     net.imglib2.RandomAccessible<T> input,
                     net.imglib2.Interval inputInterval,
                     net.imglib2.RandomAccessible<T> output,
                     net.imglib2.Localizable outputOffset,
                     net.imglib2.img.ImgFactory<T> factory,
                     T type)

Method Detail

getSigma

public double[] getSigma()

getKernel

public double[][] getKernel()

numDimensions

public int numDimensions()

getInput

public net.imglib2.RandomAccessible<T> getInput()

getOutput

public net.imglib2.RandomAccessible<T> getOutput()

getFactory

public net.imglib2.img.ImgFactory<T> getFactory()

getInputInterval

public net.imglib2.Interval getInputInterval()

getOutputOffset

public net.imglib2.Localizable getOutputOffset()

type

public T type()

getTmp1

public net.imglib2.img.Img<T> getTmp1()

getTmp2

public net.imglib2.img.Img<T> getTmp2()

getProcessingType

protected T getProcessingType()

getProcessingLine

protected abstract net.imglib2.img.Img<T> getProcessingLine(long size)

getResult

public net.imglib2.RandomAccessible<T> getResult()

Returns:

the result of the convolution

getRange

protected net.imglib2.Interval getRange(int dim)

The area for the output/temp that needs to be convolved, always relative to the input of the next convolution operation. The area is larger than the input, defined by the size of the kernel in each dimension.

Parameters:

dim - - The dimension that is currently processed

Returns:

- the Interval defining the output size for the current dimension that is processed

createInputLineSampler

protected SamplingLineIterator<T> createInputLineSampler(int dim,
                                                         net.imglib2.Interval range)

An Iterator that samples a one dimensional line of input data for the current dimension.

Parameters:

dim - - The current dimension

range - - the size of the output/temp image

Returns:

- A SamplingLineIterator which provides the input

createOutputLineWriter

protected WritableLineIterator<T> createOutputLineWriter(int dim,
                                                         net.imglib2.Interval range,
                                                         SamplingLineIterator<T> inputLineSampler)

An Iterator that samples a one dimensional line of output data for the current dimension.

Parameters:

dim - - The current dimension

range - - the size of the output/temp image

inputLineSampler - - the input line sampler which knows all the stuff already

Returns:

- A SamplingLineIterator which provides the output

processLine

protected void processLine(SamplingLineIterator<T> input,
                           double[] kernel)

Compute the current line. It is up to the implementation howto really do that. The idea is to only iterate over the input once (that's why it is an Iterator) as it is potentially an expensive operation (e.g. a Converter might be involved or we are computing on a rendered input)

Parameters:

input - - the Iterator/Sampler over the current input line.

writeLine

protected void writeLine(WritableLineIterator<T> a,
                         SamplingLineIterator<T> inputLineSampler)

Writes the computed line back into the output/temp image. The idea is to only iterate over the output once (that's why it is an Iterator) as it is potentially an expensive operation (e.g. a Converter might be involved)

Parameters:

a - - the Iterator/Sampler over the current output line.

updateInputLineSampler

protected void updateInputLineSampler(SamplingLineIterator<T> a,
                                      net.imglib2.Interval range,
                                      long[] offset,
                                      net.imglib2.Localizable originalLocation)

Updates the current SamplingLineIterator to the location of the new line that is processed and clears the processing line (set all values to zero)

Parameters:

a - - the SamplingLineIterator

range - - the size of the output/temp image

offset - - the new position

originalLocation - - the location where is was put initially, all movements are relative to that

updateOutputLineWriter

protected void updateOutputLineWriter(WritableLineIterator<T> a,
                                      net.imglib2.Interval range,
                                      long[] offset,
                                      net.imglib2.Localizable originalLocation)

Updates the current WritableLineIterator to the location of the new line that is processed.

Parameters:

a - - the WritableLineIterator

range - - the size of the output/temp image

offset - - the new position

originalLocation - - the location where is was put initially, all movements are relative to that

computeKernel

protected void computeKernel()

Compute the Gaussian Kernel for all dimensions

getTemporaryImgSize

protected net.imglib2.Interval getTemporaryImgSize()

Returns:

- the size the input images need to have

getNumThreads

public int getNumThreads()

setNumThreads

public void setNumThreads(int numThreads)

call

public void call()