lti::cwagmSegmentation Class Reference

CWAGM Segmentation. More...

#include <ltiCWAGMSegmentation.h>

Inheritance diagram for lti::cwagmSegmentation:

[legend]

Collaboration diagram for lti::cwagmSegmentation:

[legend]

List of all members.

Classes
class	parameters
	the parameters for the class cwagmSegmentation More...
Public Member Functions
	cwagmSegmentation ()
	cwagmSegmentation (const parameters &par)
	cwagmSegmentation (const cwagmSegmentation &other)
virtual	~cwagmSegmentation ()
virtual const char *	getTypeName () const
bool	apply (const image &src, imatrix &regions, ivector &regionSizes)
bool	apply (const image &src, image &preproc, imatrix &regions, ivector &regionSizes)
bool	apply (const image &src, image &preproc, imatrix &regions, ivector &regionSizes, channel &gradMag, channel &gradArg)
cwagmSegmentation &	copy (const cwagmSegmentation &other)
cwagmSegmentation &	operator= (const cwagmSegmentation &other)
virtual functor *	clone () const
const parameters &	getParameters () const
virtual bool	updateParameters ()
Protected Member Functions
bool	preprocess (const image &src, image &dest) const
bool	worker (const image &orig, const image &preproc, imatrix &regions, ivector &regionSizes, channel &gradM, channel &gradA, vector< float > &gradHisto, float &maxGrad)
bool	enhanceContrast (image &src) const
Protected Attributes
local functors

medianFilter	median
watershedSegmentation	watershed
splitImage *	splitter
colorContrastGradient	colorContrast
regionGraphColorHaris	haris

Detailed Description

CWAGM Segmentation.

CWAGM means Color Watershed - Adjacency Graph Merge. It is a segmentation algortihms based on low level information introduced in P. Alvarado "Segmentation of color images for interactive 3D object retrieval". Ph.D. Thesis. RWTH-Aachen, 2004. http://www.bth.rwth-aachen.de/job/disslist.pl

This watershed-based concept is based on the ideas of Haris et.al. for a segmentation concept of gray-valued image, extended to color images. It uses (optionally) a median filtering to reduce the noise in each color channel, computes an over-segmentation with the watershed transformation and merge the regions to minimize the mean square error of a piece-wise constant image approximation.

Constructor & Destructor Documentation

lti::cwagmSegmentation::cwagmSegmentation ( )

default constructor

lti::cwagmSegmentation::cwagmSegmentation ( const parameters & par )

Construct a functor using the given parameters.

lti::cwagmSegmentation::cwagmSegmentation ( const cwagmSegmentation & other )

copy constructor

Parameters:

other

the object to be copied

virtual lti::cwagmSegmentation::~cwagmSegmentation ( ) [virtual]

destructor

Member Function Documentation

bool lti::cwagmSegmentation::apply	(	const image &	src,
		image &	preproc,
		imatrix &	regions,
		ivector &	regionSizes,
		channel &	gradMag,
		channel &	gradArg
	)

Segmentate the image src and return many internal partial results that can be used in further processing units.

Note that there are several segmentation modes (like color ACA or mean-shift segmentation) which do not require the gradient computation. These modes will leave the given gradMag and gradArg untouched. This way, the gradients won't be unnecessarily computed and you can detect if they are computed or not: just give empty channels and if they are still empty after calling this method, you will have to compute them, if you need to.

Parameters:

	src	image with the source data.
	preproc	resulting preprocessed image.
	regions	labeled mask as imatrix, where the result will be left.
	regionSizes	sizes of each region in the regions mask.
	gradMag	magnitude of the color gradient
	gradArg	orientation of the color gradient

Returns:: true if apply successful or false otherwise.

bool lti::cwagmSegmentation::apply	(	const image &	src,
		image &	preproc,
		imatrix &	regions,
		ivector &	regionSizes
	)

operates on a copy of the given parameters.

Parameters:

	src	image with the source data.
	preproc	preprocessed image.
	regions	labeled mask as imatrix, where the result will be left.
	regionSizes	sizes of each region in the regions mask.

Returns:: true if apply successful or false otherwise.

bool lti::cwagmSegmentation::apply	(	const image &	src,
		imatrix &	regions,
		ivector &	regionSizes
	)

operates on a copy of the given parameters.

Parameters:

	src	image with the source data.
	regions	labeled mask as imatrix, where the result will be left.
	regionSizes	sizes of each region in the regions mask.

Returns:: true if apply successful or false otherwise.

virtual functor* lti::cwagmSegmentation::clone ( ) const [virtual]

returns a pointer to a clone of this functor.

Reimplemented from lti::segmentation.

cwagmSegmentation& lti::cwagmSegmentation::copy ( const cwagmSegmentation & other )

copy data of "other" functor.

Parameters:

other

the functor to be copied

Returns:: a reference to this functor object

Reimplemented from lti::segmentation.

bool lti::cwagmSegmentation::enhanceContrast ( image & src ) const [protected]

The simplest contrast enhancement consists in mapping the input value range of each channel into 0 to 255, each channel independently of the others.

src MUST be connected. (see matrix<T>::mode())

const parameters& lti::cwagmSegmentation::getParameters ( ) const

returns used parameters

Reimplemented from lti::functor.

virtual const char* lti::cwagmSegmentation::getTypeName ( ) const [virtual]

returns the name of this type ("cwagmSegmentation")

Reimplemented from lti::segmentation.

cwagmSegmentation& lti::cwagmSegmentation::operator= ( const cwagmSegmentation & other )

alias for copy member

Parameters:

other

the functor to be copied

Returns:: a reference to this functor object

Reimplemented from lti::functor.

bool lti::cwagmSegmentation::preprocess	(	const image &	src,
		image &	dest
	)			const `[protected]`

Preprocess Denoise and Enhance Contrast.

virtual bool lti::cwagmSegmentation::updateParameters ( ) [virtual]

Update functor's parameters.

Update the internal state of the functor according to the parameters. This special case the functor used to split the image into its color componentes is initialized here.

Returns:: true if successful, false otherwise

Reimplemented from lti::functor.

bool lti::cwagmSegmentation::worker	(	const image &	orig,
		const image &	preproc,
		imatrix &	regions,
		ivector &	regionSizes,
		channel &	gradM,
		channel &	gradA,
		vector< float > &	gradHisto,
		float &	maxGrad
	)			`[protected]`

Split and merge with watersheds.

Parameters:

	orig	original color image
	preproc	preprocessed image
	regions	labeld mask with initial partition
	regionSizes	number of pixels per region
	gradM	magnitude of the color gradient
	gradA	orientation of the color gradient
	gradHisto	histogram of the gradient values
	maxGrad	maximum gradient value which is also the last value in the histogram.