A tool to separate a complex image into individual components. More...

#include <ImageDecomposer.h>

Public Types
enum	componentValues { INDETERMINATE, MASKED }
	'Special' flag values for pixels in the component map. More...
Public Member Functions
	ImageDecomposer ()
	Default constructor.
	ImageDecomposer (ImageInterface< T > &image)
	Construct from image.
	ImageDecomposer (const ImageDecomposer< T > &other)
	Copy constructor.
ImageDecomposer< T > &	operator= (const ImageDecomposer< T > &other)
	Assignment.
	~ImageDecomposer ()
	Destructor.
void	setImage (ImageInterface< T > &image)
	Tell the decomposer what image to decompose ("target image").
void	setDeblend (Bool deblendIt=True)
	Tells the program whether or not to use the contour-based deblender.
void	setDeblendOptions (T thresholdVal=0.1, uInt nContour=11, Int minRange=2, Int nAxis=2)
	Specifies deblending options:
void	setFit (Bool fitIt=True)
	Tells the program whether or not to perform fitting.
void	setFitOptions (T maximumRMS=0.1, Int maxRetries=-1, uInt maxIter=256, T convCriteria=0.0001)
	Specifies fitting options:
void	decomposeImage ()
	The primary method of this class - executes the instructions stated in the options above by deblending and/or fitting to the image to generate the component map and/or component list.
uInt	numRegions () const
	Returns the number of regions found in the image.
uInt	numComponents () const
	Returns the number of components found in the image.
IPosition	shape () const
	Returns the shape of the component map.
Int	shape (uInt axis) const
	Returns the length of a specific axis.
Bool	isDerived () const
	Returns True if the image has been thresholded (split up into regions.)
Bool	isDecomposed () const
	Returns True if the image has been decomposed (split up into components.)
Matrix< T >	componentList () const
	Returns the component parameters as a Matrix.
void	componentMap () const
	Currently does nothing; in the future should return the component map in a way that it can be seen by the user in AIPS++, preferably as a colorized image.
void	display () const
	Command-line text output functions.
void	displayContourMap (const Vector< T > &clevels) const
void	printComponents () const
void	boundRegions (Block< IPosition > &blc, Block< IPosition > &trc)
	Boxes each region in the componentmap: blc is set to the lowest coordinate value in each region; trc is set to one above the highest coordinate value in each region.
Private Member Functions
void	copyOptions (const ImageDecomposer< T > &other)
void	correctBlcTrc (IPosition &blc, IPosition &trc) const
	Makes sure a pair of IPositions is in the correct format for blc/trc, and corrects them if they are not.
Bool	increment (IPosition &pos, const IPosition &shape) const
	Used as an N-dimensional interator.
void	decrement (IPosition &pos) const
Int	getCell (Int x, Int y) const
	Returns the component to which the specified cell belongs.
Int	getCell (Int x, Int y, Int z) const
Int	getCell (const IPosition &coord) const
void	setCell (Int x, Int y, Int sval)
	Assigns the specified cell to the specified component.
void	setCell (Int x, Int y, Int z, Int sval)
void	setCell (const IPosition &coord, Int sval)
Vector< T >	autoContour (T minCon, T maxCon, T inc) const
	Semi-automatic way to set contour levels: at the given increment counting between mincon and maxcon.
Vector< T >	autoContour (Int nContours=11, T minValue=0) const
	Linearly spaces contours between minvalue and just below the maximum value in the target region of the target image, and returns the contour values as a Vector.
Vector< T >	autoContour (const Function1D< T > &fn, Int nContours=11, T minValue=0) const
	Nonlinear spacing option for contouring; spaces contours according to the function given.
void	destroyRegions (const Vector< Bool > &killRegion)
	Eliminates any regions whose corresponding values in killRegion are True by setting all pixel values in the componentmap set to that region to zero.
void	renumberRegions ()
	Eliminates regions with no cells by replacing them with higher-numbered regions.
void	synthesize (const ImageDecomposer< T > &subdecomposer, IPosition blc)
	Overlays a smaller map onto an empty region of a larger map, and adds submap component list to main component list.
void	zero ()
	Set all elements in the component map to zero and clear the component list.
void	clear ()
	Set all nonmasked elements in the component map to zero and clear the component list.
T	findAreaGlobalMax (IPosition blc, IPosition trc) const
	Finds the greatest value inside the specified rectangular area of the target image.
void	findAreaGlobalMax (T &maxval, IPosition &maxvalpos, IPosition blc, IPosition trc) const
Vector< T >	findAreaGlobalMax (IPosition blc, IPosition trc, Int naxis) const
void	findAreaGlobalMax (Vector< T > &maxvals, Block< IPosition > &maxvalpos, IPosition blc, IPosition trc, Int naxis) const
Vector< T >	findAreaLocalMax (IPosition blc, IPosition trc, Int naxis) const
	Finds all local maxima inside the specified rectangular area of the target image.
void	findAreaLocalMax (Vector< T > &maxvals, Block< IPosition > &maxvalpos, IPosition blc, IPosition trc, Int naxis) const
Vector< T >	findAllRegionGlobalMax () const
	Finds the maximum value of the target image in each region of the componentmap.
void	findAllRegionGlobalMax (Vector< T > &maxvals, Block< IPosition > &maxvalpos) const
Vector< T >	findRegionLocalMax (Int nregion, Int naxis) const
	Finds all local maxima of the target image inside the specifed region of the componentmap.
void	findRegionLocalMax (Vector< T > &maxvals, Block< IPosition > &maxvalpos, Int nregion, Int naxis) const
Bool	isLocalMax (const IPosition &pos, Int naxis) const
	Compares specified pixel to adjacent pixels to determine if it is greatest in local pixel block.
Bool	isLocalMax (Int x, Int y, Int naxis) const
Bool	isLocalMax (Int x, Int y, Int z, Int naxis) const
void	estimateComponentWidths (Matrix< T > &width, const Block< IPosition > &maxvalpos) const
	Finds a rough estimate of the width of each component by scanning to find the full width at quarter maximum.
Array< T >	calculateMoments (Int region) const
	Calculates the 0th-2nd order moments of a region.
uInt	identifyRegions (T thrval, Int naxis=2)
	Performs a single threshold scan on the image.
void	deblendRegions (const Vector< T > &contours, Int minRange=1, Int naxis=2)
	Performs the contour decomposition on a blended image to generate a component map that can detect components blended above any threshold(s), by performing threshold scans at each contour level and recognizing as individual any components that are distinct above any such level.
T	getImageVal (IPosition coord) const
	Retrieves the target image's value at the given location.
T	getImageVal (Int x, Int y) const
T	getImageVal (Int x, Int y, Int z) const
Int	getContourVal (IPosition coord, const Vector< T > &clevels) const
	Retrieves the number of the highest contour with a value less then the target image's value at the given location.
Int	getContourVal (Int x, Int y, Int z, const Vector< T > &clevels) const
Int	getContourVal (Int x, Int y, const Vector< T > &clevels) const
Int	getContourVal (T val, const Vector< T > &clevels) const
Matrix< T >	fitRegion (Int region)
	Fits multiple gaussians to a single region.
void	fitRegions ()
	Fits gaussians to an image; multiple gaussians per region in the component map.
void	fitComponents ()
	Fits gaussians to an image; one gaussian per region in the pmap.
Matrix< T >	estimateComponents ()
	Estimate the component parameters based on moments calculated using the component map.
Matrix< T >	fitGauss (const Matrix< T > &positions, const Vector< T > &dataValues, const Matrix< T > &initestimate) const
	Fits the specified number of 3D gaussians to the data, and returns solution in image (world) coordinates.
Private Attributes
ImageInterface< T > *	itsImagePtr
Lattice< Int > *	itsMapPtr
IPosition	itsShape
uInt	itsDim
uInt	itsNRegions
uInt	itsNComponents
Matrix< T >	itsList
Bool	itsDeblendIt
	each component.)
T	itsThresholdVal
uInt	itsNContour
Int	itsMinRange
Int	itsNAxis
Bool	itsFitIt
T	itsMaximumRMS
Int	itsMaxRetries
uInt	itsMaxIter
T	itsConvCriteria

Detailed Description

template<class T>
class casa::ImageDecomposer< T >

A tool to separate a complex image into individual components.

Intended use:

Public interface

Review Status

Test programs:: tImageDecomposer

Prerequisite

Etymology

It takes an image, and separates it into components.

Synopsis

ImageDecomposer is an image decomposition tool that performs several tasks, with the end result being that a strongly blended image is separated into components - both in the sense that it determines the parameters for each component (assuming a Gaussian model) and that it physically assigns each pixel in the image to an individual object. The products of these two operations are called the component list and the component map, respectively. The fitting process (which determines the component list) and the pixel-decomposition process (which determines the component map) are designed to work cooperatively to increase the efficiency and accuracy of both, though each can operate without the other if necessary.

The algorithm between the decomposition is based on the function clfind described in Williams et al 1994, which uses a contouring procedure whereby a closed contour designates a separate component. The program first separates the image into clearly distint 'regions' of blended emission, then contours each region to determine the areas constituting each component and passes this information on to the fitter, which determines the component list.

The software is compatible with 2 and 3 dimensional images, but is not yet structured for higher dimensions.

Example

     TempImage<Double> image;
     //(populate the image with data: see dImageDecomposer.cc)
     ImageDecomposer<Double> id(image);
     id.setDeblendOptions(0.3, 8);
     id.setFitOptions(0.4);
     id.decomposeImage();
     id.display();
     id.printComponents();

Motivation

Note:

To Do

Generalize dimensionality
Use Lattice iterators in place of IPosition loops wherever possible
Speed up fitting by not sending every region pixel to the fitter
Send the completed componentmap to the user as an AIPS++ (Int?) Image
Return a ComponentList instead of a Matrix
Enable custom contouring at user level
Add progress meter
Numerous other improvements to make are documented in the code

Definition at line 119 of file ImageDecomposer.h.

Member Enumeration Documentation

template<class T>

enum casa::ImageDecomposer::componentValues

'Special' flag values for pixels in the component map.

An indeterminate pixel lies directly between two components and cannot be immediately assigned. A masked pixel is not inside the targeted region of the sub-componentmap and is not used in decomposition or fitting.

Enumerator:

INDETERMINATE
MASKED

Definition at line 127 of file ImageDecomposer.h.

Constructor & Destructor Documentation

template<class T>

casa::ImageDecomposer< T >::ImageDecomposer ( )

Default constructor.

Object is not viable until setImage called

template<class T>

casa::ImageDecomposer< T >::ImageDecomposer ( ImageInterface< T > & image )

Construct from image.

template<class T>

casa::ImageDecomposer< T >::ImageDecomposer ( const ImageDecomposer< T > & other )

Copy constructor.

template<class T>

casa::ImageDecomposer< T >::~ImageDecomposer ( )

Destructor.

Member Function Documentation

template<class T>

Vector<T> casa::ImageDecomposer< T >::autoContour	(	T	minCon,
		T	maxCon,
		T	inc
	)		const `[private]`

Semi-automatic way to set contour levels: at the given increment counting between mincon and maxcon.

template<class T>

Vector<T> casa::ImageDecomposer< T >::autoContour	(	Int	nContours = `11`,
		T	minValue = `0`
	)		const `[private]`

Linearly spaces contours between minvalue and just below the maximum value in the target region of the target image, and returns the contour values as a Vector.

template<class T>

Vector<T> casa::ImageDecomposer< T >::autoContour	(	const Function1D< T > &	fn,
		Int	nContours = `11`,
		T	minValue = `0`
	)		const `[private]`

Nonlinear spacing option for contouring; spaces contours according to the function given.

The domain of the function is 0 <-> ncontours-1; the range is automatically calibrated to be minvalue <-> maxvalue. The function should be nondecreasing in the domain such that each contour is greater than the last.

template<class T>

void casa::ImageDecomposer< T >::boundRegions	(	Block< IPosition > &	blc,
		Block< IPosition > &	trc
	)

Boxes each region in the componentmap: blc is set to the lowest coordinate value in each region; trc is set to one above the highest coordinate value in each region.

template<class T>

Array<T> casa::ImageDecomposer< T >::calculateMoments ( Int region ) const [private]

Calculates the 0th-2nd order moments of a region.

template<class T>

void casa::ImageDecomposer< T >::clear ( ) [private]

Set all nonmasked elements in the component map to zero and clear the component list.

template<class T>

Matrix<T> casa::ImageDecomposer< T >::componentList ( ) const

Returns the component parameters as a Matrix.

(Ideally, this should be a ComponentList.)

template<class T>

void casa::ImageDecomposer< T >::componentMap ( ) const

Currently does nothing; in the future should return the component map in a way that it can be seen by the user in AIPS++, preferably as a colorized image.

template<class T>

void casa::ImageDecomposer< T >::copyOptions ( const ImageDecomposer< T > & other ) [private]

template<class T>

void casa::ImageDecomposer< T >::correctBlcTrc	(	IPosition &	blc,
		IPosition &	trc
	)		const `[private]`

Makes sure a pair of IPositions is in the correct format for blc/trc, and corrects them if they are not.

template<class T>

void casa::ImageDecomposer< T >::deblendRegions	(	const Vector< T > &	contours,
		Int	minRange = `1`,
		Int	naxis = `2`
	)		`[private]`

Performs the contour decomposition on a blended image to generate a component map that can detect components blended above any threshold(s), by performing threshold scans at each contour level and recognizing as individual any components that are distinct above any such level.

template<class T>

void casa::ImageDecomposer< T >::decomposeImage ( )

The primary method of this class - executes the instructions stated in the options above by deblending and/or fitting to the image to generate the component map and/or component list.

template<class T>

void casa::ImageDecomposer< T >::decrement ( IPosition & pos ) const [private]

template<class T>

void casa::ImageDecomposer< T >::destroyRegions ( const Vector< Bool > & killRegion ) [private]

Eliminates any regions whose corresponding values in killRegion are True by setting all pixel values in the componentmap set to that region to zero.

Zero-oriented; there is an offset of one between the index in killRegion and the actual region in the componentmap.

template<class T>

void casa::ImageDecomposer< T >::display ( ) const

Command-line text output functions.

template<class T>

void casa::ImageDecomposer< T >::displayContourMap ( const Vector< T > & clevels ) const

template<class T>

Matrix<T> casa::ImageDecomposer< T >::estimateComponents ( ) [private]

Estimate the component parameters based on moments calculated using the component map.

template<class T>

void casa::ImageDecomposer< T >::estimateComponentWidths	(	Matrix< T > &	width,
		const Block< IPosition > &	maxvalpos
	)		const `[private]`

Finds a rough estimate of the width of each component by scanning to find the full width at quarter maximum.

Requires the location of each component. This function is mostly obsolete, and is only used when the contour deblender is off (since the component map is necessary to determine the moments).

template<class T>

Vector<T> casa::ImageDecomposer< T >::findAllRegionGlobalMax ( ) const [private]

Finds the maximum value of the target image in each region of the componentmap.

template<class T>

void casa::ImageDecomposer< T >::findAllRegionGlobalMax	(	Vector< T > &	maxvals,
		Block< IPosition > &	maxvalpos
	)		const `[private]`

template<class T>

T casa::ImageDecomposer< T >::findAreaGlobalMax	(	IPosition	blc,
		IPosition	trc
	)		const `[private]`

Finds the greatest value inside the specified rectangular area of the target image.

template<class T>

void casa::ImageDecomposer< T >::findAreaGlobalMax	(	T &	maxval,
		IPosition &	maxvalpos,
		IPosition	blc,
		IPosition	trc
	)		const `[private]`

template<class T>

Vector<T> casa::ImageDecomposer< T >::findAreaGlobalMax	(	IPosition	blc,
		IPosition	trc,
		Int	naxis
	)		const `[private]`

template<class T>

void casa::ImageDecomposer< T >::findAreaGlobalMax	(	Vector< T > &	maxvals,
		Block< IPosition > &	maxvalpos,
		IPosition	blc,
		IPosition	trc,
		Int	naxis
	)		const `[private]`

template<class T>

Vector<T> casa::ImageDecomposer< T >::findAreaLocalMax	(	IPosition	blc,
		IPosition	trc,
		Int	naxis
	)		const `[private]`

Finds all local maxima inside the specified rectangular area of the target image.

template<class T>

void casa::ImageDecomposer< T >::findAreaLocalMax	(	Vector< T > &	maxvals,
		Block< IPosition > &	maxvalpos,
		IPosition	blc,
		IPosition	trc,
		Int	naxis
	)		const `[private]`

template<class T>

Vector<T> casa::ImageDecomposer< T >::findRegionLocalMax	(	Int	nregion,
		Int	naxis
	)		const `[private]`

Finds all local maxima of the target image inside the specifed region of the componentmap.

template<class T>

void casa::ImageDecomposer< T >::findRegionLocalMax	(	Vector< T > &	maxvals,
		Block< IPosition > &	maxvalpos,
		Int	nregion,
		Int	naxis
	)		const `[private]`

template<class T>

void casa::ImageDecomposer< T >::fitComponents ( ) [private]

Fits gaussians to an image; one gaussian per region in the pmap.

This function is intended to be used only by ImageDecomposer on its intermediary subimages; using it at higher level will execute a full gaussian fit on the main image and will be extremely slow. Every nonflagged object pixel in the image is used in fitting.

If the deblended flag is True, the function will treat each region as an individual component and will fit that many gaussians to the image

template<class T>

Matrix<T> casa::ImageDecomposer< T >::fitGauss	(	const Matrix< T > &	positions,
		const Vector< T > &	dataValues,
		const Matrix< T > &	initestimate
	)		const `[private]`

Fits the specified number of 3D gaussians to the data, and returns solution in image (world) coordinates.

Essentially just an interface for FitGaussian.

template<class T>

Matrix<T> casa::ImageDecomposer< T >::fitRegion ( Int region ) [private]

Fits multiple gaussians to a single region.

First performs a local maximum scan to estimate the number of components in the region.

template<class T>

void casa::ImageDecomposer< T >::fitRegions ( ) [private]

Fits gaussians to an image; multiple gaussians per region in the component map.

The regions are fit sequentially and independently, so this function can be used on the main image. If the map is not yet thresholded, will fit to the entire image as if it were a single composite object, which will be very slow.

template<class T>

Int casa::ImageDecomposer< T >::getCell	(	Int	x,
		Int	y
	)		const `[private]`

Returns the component to which the specified cell belongs.

template<class T>

Int casa::ImageDecomposer< T >::getCell	(	Int	x,
		Int	y,
		Int	z
	)		const `[private]`

template<class T>

Int casa::ImageDecomposer< T >::getCell ( const IPosition & coord ) const [private]

template<class T>

Int casa::ImageDecomposer< T >::getContourVal	(	IPosition	coord,
		const Vector< T > &	clevels
	)		const `[private]`

Retrieves the number of the highest contour with a value less then the target image's value at the given location.

template<class T>

Int casa::ImageDecomposer< T >::getContourVal	(	Int	x,
		Int	y,
		Int	z,
		const Vector< T > &	clevels
	)		const `[private]`

template<class T>

Int casa::ImageDecomposer< T >::getContourVal	(	Int	x,
		Int	y,
		const Vector< T > &	clevels
	)		const `[private]`

template<class T>

Int casa::ImageDecomposer< T >::getContourVal	(	T	val,
		const Vector< T > &	clevels
	)		const `[private]`

template<class T>

T casa::ImageDecomposer< T >::getImageVal ( IPosition coord ) const [private]

Retrieves the target image's value at the given location.

template<class T>

T casa::ImageDecomposer< T >::getImageVal	(	Int	x,
		Int	y
	)		const `[private]`

template<class T>

T casa::ImageDecomposer< T >::getImageVal	(	Int	x,
		Int	y,
		Int	z
	)		const `[private]`

template<class T>

uInt casa::ImageDecomposer< T >::identifyRegions	(	T	thrval,
		Int	naxis = `2`
	)		`[private]`

Performs a single threshold scan on the image.

In other words, identifies all contigous blocks of pixels in the target image above the threshold value thrval, assigning each unique block to an integer, starting at one. All pixels with target image values below thrval are set to zero.

template<class T>

Bool casa::ImageDecomposer< T >::increment	(	IPosition &	pos,
		const IPosition &	shape
	)		const `[private]`

Used as an N-dimensional interator.

This should probably be replaced by LatticeIterators...?

template<class T>

Bool casa::ImageDecomposer< T >::isDecomposed ( ) const

Returns True if the image has been decomposed (split up into components.)

template<class T>

Bool casa::ImageDecomposer< T >::isDerived ( ) const

Returns True if the image has been thresholded (split up into regions.)

template<class T>

Bool casa::ImageDecomposer< T >::isLocalMax	(	const IPosition &	pos,
		Int	naxis
	)		const `[private]`

Compares specified pixel to adjacent pixels to determine if it is greatest in local pixel block.

2D: naxis = 1: compare to 4 adjacent pixels (axes only) naxis = 2: compare to 8 adjacent pixels (axes and diagonals) 3D: naxis = 1: compare to 6 adjacent pixels (axes only) naxis = 2: compare to 18 adjacent pixels (axes and 2-axis diagonals) naxis = 3: compare to 26 adjacent pixels (axes and 2/3-axis diagonals)

template<class T>

Bool casa::ImageDecomposer< T >::isLocalMax	(	Int	x,
		Int	y,
		Int	naxis
	)		const `[private]`

template<class T>

Bool casa::ImageDecomposer< T >::isLocalMax	(	Int	x,
		Int	y,
		Int	z,
		Int	naxis
	)		const `[private]`

template<class T>

uInt casa::ImageDecomposer< T >::numComponents ( ) const

Returns the number of components found in the image.

A 'component' as defined in this code is a source that can be described as a single Gaussian. This can only be determined after deblending.

template<class T>

uInt casa::ImageDecomposer< T >::numRegions ( ) const

Returns the number of regions found in the image.

A 'region' as defined in this code is a subset of the image of contiguous pixels whose values are greater than the threshold value specified in decomposeImage. A region may contain one or more components.

template<class T>

ImageDecomposer<T>& casa::ImageDecomposer< T >::operator= ( const ImageDecomposer< T > & other )

Assignment.

template<class T>

void casa::ImageDecomposer< T >::printComponents ( ) const

template<class T>

void casa::ImageDecomposer< T >::renumberRegions ( ) [private]

Eliminates regions with no cells by replacing them with higher-numbered regions.

template<class T>

void casa::ImageDecomposer< T >::setCell	(	Int	x,
		Int	y,
		Int	sval
	)		`[private]`

Assigns the specified cell to the specified component.

template<class T>

void casa::ImageDecomposer< T >::setCell	(	Int	x,
		Int	y,
		Int	z,
		Int	sval
	)		`[private]`

template<class T>

void casa::ImageDecomposer< T >::setCell	(	const IPosition &	coord,
		Int	sval
	)		`[private]`

template<class T>

void casa::ImageDecomposer< T >::setDeblend ( Bool deblendIt = True )

Tells the program whether or not to use the contour-based deblender.

If not, the program will instead perform a single thresholding followed by a local maximum scan before fitting.

template<class T>

void casa::ImageDecomposer< T >::setDeblendOptions	(	T	thresholdVal = `0.1`,
		uInt	nContour = `11`,
		Int	minRange = `2`,
		Int	nAxis = `2`
	)

Specifies deblending options:

thresholdVal: noise cutoff level, used to distinguish source pixels from background pixels. Also, regions which are not blended above this value will be fit separately.
nContour: number of total contours to use in deblending regions.
minRange: the minimum number of contours necessary to distinguish an object as a separate component.
nAxis: paramater used to define whether or not two adjacent blocks of pixels are contiguous - see identifyRegions for details.

See decomposeImage for more information on the deblending process.

template<class T>

void casa::ImageDecomposer< T >::setFit ( Bool fitIt = True )

Tells the program whether or not to perform fitting.

If not, the component list will be dstermined by estimation from the values of the first and second order moments of each component.

template<class T>

void casa::ImageDecomposer< T >::setFitOptions	(	T	maximumRMS = `0.1`,
		Int	maxRetries = `-1`,
		uInt	maxIter = `256`,
		T	convCriteria = `0.0001`
	)

Specifies fitting options:

maximumRMS: The maximum RMS residual value (after fitting) allowed to identify a fit as successful.
maxRetries: the maximum number of times the fit will be restarted in order to try to reach a successful result (convergent with RMS < maximumRMS). The default value of -1 tells the program to calculate a reasonable value automatically based on the complexity of the image.
maxIter: maximum number of iterations to spend on each fit.
convCriteria: criterion to establish convergence: see NonLinearFitLM.

Additional information on these parameters can be found in FitGaussian.

template<class T>

void casa::ImageDecomposer< T >::setImage ( ImageInterface< T > & image )

Tell the decomposer what image to decompose ("target image").

Also resets the internal component map.

template<class T>

IPosition casa::ImageDecomposer< T >::shape ( ) const

Returns the shape of the component map.

template<class T>

Int casa::ImageDecomposer< T >::shape ( uInt axis ) const

Returns the length of a specific axis.

template<class T>

void casa::ImageDecomposer< T >::synthesize	(	const ImageDecomposer< T > &	subdecomposer,
		IPosition	blc
	)		`[private]`

Overlays a smaller map onto an empty region of a larger map, and adds submap component list to main component list.

The user should exercise caution with this function and synthesize submaps only into regions of the main map that are truly empty (0), as no blending is assumed between different maps.

template<class T>

void casa::ImageDecomposer< T >::zero ( ) [private]

Set all elements in the component map to zero and clear the component list.