Detailed Description

Version:: 0.1

Author:: ayman.zureiki@laas.fr

This module implements the Stereo Correlation Algorithm. Originally, it was in calife StereoCorrelation module, but it is heavilly rewritten.

History

0.1 (2005-04-13) - Initial version

Macro

Stereo.tcl defines a macro that uses tow stereo images to produce an image of 3D points

Tcl interface (generated by swig)

The interface of the module is generated from the following files:

stereo.i defines the wrapped classes and functions,
stereoException.i defines the try { } catch block for this module.

Classes
class	jafar::stereo::CorrelationParams
	Contains all parameters needed for Rectification. More...
class	jafar::stereo::CorrelationResults
	Contains the result of Correlation. More...
class	jafar::stereo::ImagesSizes
	In this class we save all the values of the sizes of differents images. More...
class	jafar::stereo::ImagesTemporary
	Contains temporary buffers and variable needed in different algorithm. More...
class	jafar::stereo::ReconstructionParams
	Contains all parameters needed for Reconstruction 3D. More...
class	jafar::stereo::RectificationParams
	Contains all parameters needed for Rectification. More...
class	jafar::stereo::StereoException
	Base class for all exceptions defined in the module stereo. More...
Functions
jafar::image::Image *	jafar::stereo::CorrelationResults::getImageForDisparity ()
	convert the disparity image in the CorrelationResult into a Image (for display purposes)
jafar::image::Image *	jafar::stereo::CorrelationResults::getImageOfCorrelatedPixels (jafar::image::Image &rectifiedImage)
	Get a JarImage as a copy of the rectified image and in which the value of not correlated pixels (disparity == 0) are set to 0 (for display purposes)
::jafar::image::Image *	jafar::stereo::getImageForDisparity (CorrelationResults &correlationResults)
	convert the disparity image in the CorrelationResult into a Image (for display purposes)
void	jafar::stereo::GraphCutFULL (jafar::image::Image &rectifiedImage_g, jafar::image::Image &rectifiedImage_d, int nMaxDisparity, int Upq_max, char *disparityFile)
	Stereo matching using graph cut.
jafar::image::Image *	jafar::stereo::ImagesTemporary::getImageForMoyennes ()
	convert the moyenne image in the CorrelationResult into a Image (for display purposes)
void	jafar::stereo::jfrBlobFilterDisparites (CorrelationResults &correlationResults, int sizeMin, float diffMax)
	Blob_Filter.
void	jafar::stereo::jfrCorrelationCensus (::jafar::image::Image &srcImage_gauche,::jafar::image::Image &srcImage_droite, ImagesSizes &imagesSizes, CorrelationParams &correlationParams, CorrelationResults &correlationResults)
	Correlation_Census.
void	jafar::stereo::jfrCorrelationZncc (::jafar::image::Image &srcImage_gauche,::jafar::image::Image &srcImage_droite, ImagesTemporary &images_tmp_gauches, ImagesTemporary &images_tmp_droites, CorrelationParams &correlationParams, CorrelationResults &correlationResults)
	Correlation_Census.
void	jafar::stereo::jfrTransform_census_8bits (::jafar::image::Image &srcImage, ImagesTemporary &imagesTmp, CorrelationParams::CENSUS_TYPE type_census,::jafar::image::Image &imCensus)
	Transform_census.
void	jafar::stereo::jfrMoyennes (::jafar::image::Image &srcImage, int win_nb_lig, int win_nb_col, ImagesTemporary &imagesTmp)
	calculates the Mean Image using a window of given size.
void	jafar::stereo::jfrMoyennesSigmas (::jafar::image::Image &srcImage, ImagesTemporary &imagesTmp, CorrelationParams &correlationParams)
	calculates the Mean Image and covariance using given correlation parameters.
void	jafar::stereo::jfrReconstruction3D (CorrelationResults &correlationResults, ImagesSizes &imagesSizes, ReconstructionParams::RECONSTRUCTION_REPERE type_repere, ReconstructionParams::RECONSTRUCTION_TYPE type_recontruction, ReconstructionParams &reconstructionParams, RectificationParams &rectificationParamsGauche, model3d::Image3DTemplate< model3d::Point3D > &im3D)
	implements the Reconstruction algorithm.
void	jafar::stereo::jfrReconstruction3DVar (CorrelationResults &correlationResults, ImagesSizes &imagesSizes, ReconstructionParams::RECONSTRUCTION_REPERE type_repere, ReconstructionParams::RECONSTRUCTION_TYPE type_recontruction, ReconstructionParams &reconstructionParams, RectificationParams &rectificationParamsGauche, model3d::Image3DTemplate< model3d::Point3DVar > &im3DVar)
	implements the Reconstruction algorithm and calculates errors.
void	jafar::stereo::jfrRectification (::jafar::image::Image &srcImage,::jafar::image::Image &destImage, ImagesSizes &imagesSizes, RectificationParams &rectifParams, int reduction, RectificationParams::RECTIFICATION_TYPE type_rectification, RectificationParams::DISTORTION_TYPE type_distortion, RectificationParams::DO_INTERPOLATION do_interpolation)
	implements the Rectifiction algorithm

Function Documentation

::jafar::image::Image* jafar::stereo::getImageForDisparity ( CorrelationResults & correlationResults )

convert the disparity image in the CorrelationResult into a Image (for display purposes)

Returns:: Image: an image for the disparity

jafar::image::Image* jafar::stereo::CorrelationResults::getImageForDisparity ( )

convert the disparity image in the CorrelationResult into a Image (for display purposes)

Returns:: Image: an image for the disparity

jafar::image::Image* jafar::stereo::ImagesTemporary::getImageForMoyennes ( )

convert the moyenne image in the CorrelationResult into a Image (for display purposes)

Returns:: Image: an image for the moyonne image

jafar::image::Image* jafar::stereo::CorrelationResults::getImageOfCorrelatedPixels ( jafar::image::Image & rectifiedImage )

Get a JarImage as a copy of the rectified image and in which the value of not correlated pixels (disparity == 0) are set to 0 (for display purposes)

Returns:: Image: an image for the disparity

void jafar::stereo::GraphCutFULL	(	jafar::image::Image &	rectifiedImage_g,
		jafar::image::Image &	rectifiedImage_d,
		int	nMaxDisparity,
		int	Upq_max,
		char *	disparityFile
	)

Stereo matching using graph cut.

The graph is a complete one, each node has six neibghours, and the the graph has nMaxDisparity layers. The images must be rectified.

Parameters:

rectifiedImage_g,rectifiedImage_g	: the rectified left and right images.
nMaxDisparity	: the maximum value of disparity. It must be positive. we suppose the range of disparity [0, nMaxDisparity(
Upq_max	: This is a smoothing parameter. It defines the maximum value of the smoothin function u(p,q), where p and q are tow adjacent pixels. the value 100 may be a good start point for tuning.
disparityFile	: a file name where the resulting disparity image will be saved. you must give the extension of the file also, for example, "disparity.ppm".

Note:: : This is algorithm is very hungry for memory. It is recommanded to not use it with image bigger than 128*128 with MaxMisparity bigger than 20.

void jafar::stereo::jfrBlobFilterDisparites	(	CorrelationResults &	correlationResults,
		int	sizeMin,
		float	diffMax
	)

Blob_Filter.

Note:: All parameters must be reserved before calling this function

void jafar::stereo::jfrCorrelationCensus	(	::jafar::image::Image &	srcImage_gauche,
		::jafar::image::Image &	srcImage_droite,
		ImagesSizes &	imagesSizes,
		CorrelationParams &	correlationParams,
		CorrelationResults &	correlationResults
	)

Correlation_Census.

Note:: All parameters must be reserved before calling this function

void jafar::stereo::jfrCorrelationZncc	(	::jafar::image::Image &	srcImage_gauche,
		::jafar::image::Image &	srcImage_droite,
		ImagesTemporary &	images_tmp_gauches,
		ImagesTemporary &	images_tmp_droites,
		CorrelationParams &	correlationParams,
		CorrelationResults &	correlationResults
	)

Correlation_Census.

Note:: All parameters must be reserved before calling this function

void jafar::stereo::jfrMoyennes	(	::jafar::image::Image &	srcImage,
		int	win_nb_lig,
		int	win_nb_col,
		ImagesTemporary &	imagesTmp
	)

calculates the Mean Image using a window of given size.

Note:: The distination image must be reserved before calling this function

void jafar::stereo::jfrMoyennesSigmas	(	::jafar::image::Image &	srcImage,
		ImagesTemporary &	imagesTmp,
		CorrelationParams &	correlationParams
	)

calculates the Mean Image and covariance using given correlation parameters.

Note:: The distination image must be reserved before calling this function

void jafar::stereo::jfrReconstruction3D	(	CorrelationResults &	correlationResults,
		ImagesSizes &	imagesSizes,
		ReconstructionParams::RECONSTRUCTION_REPERE	type_repere,
		ReconstructionParams::RECONSTRUCTION_TYPE	type_recontruction,
		ReconstructionParams &	reconstructionParams,
		RectificationParams &	rectificationParamsGauche,
		model3d::Image3DTemplate< model3d::Point3D > &	im3D
	)

implements the Reconstruction algorithm.

Note:: All parameters must be reserved and initialized before passing into this function

Returns:: Image3D: an image of 3D points

void jafar::stereo::jfrReconstruction3DVar	(	CorrelationResults &	correlationResults,
		ImagesSizes &	imagesSizes,
		ReconstructionParams::RECONSTRUCTION_REPERE	type_repere,
		ReconstructionParams::RECONSTRUCTION_TYPE	type_recontruction,
		ReconstructionParams &	reconstructionParams,
		RectificationParams &	rectificationParamsGauche,
		model3d::Image3DTemplate< model3d::Point3DVar > &	im3DVar
	)

implements the Reconstruction algorithm and calculates errors.

Note:: All parameters must be reserved and initialized before passing into this function

Returns:: Image3DVar: an image of 3D points with covariance matrix

void jafar::stereo::jfrRectification	(	::jafar::image::Image &	srcImage,
		::jafar::image::Image &	destImage,
		ImagesSizes &	imagesSizes,
		RectificationParams &	rectifParams,
		int	reduction,
		RectificationParams::RECTIFICATION_TYPE	type_rectification,
		RectificationParams::DISTORTION_TYPE	type_distortion,
		RectificationParams::DO_INTERPOLATION	do_interpolation
	)

implements the Rectifiction algorithm

Note:: the destination image must be reserved before calling this function; Do Attention: here we do the rectification and the reduction in one step

void jafar::stereo::jfrTransform_census_8bits	(	::jafar::image::Image &	srcImage,
		ImagesTemporary &	imagesTmp,
		CorrelationParams::CENSUS_TYPE	type_census,
		::jafar::image::Image &	imCensus
	)

Transform_census.

Note:: All parameters must be reserved before calling this function

Detailed Description

History

Macro

Tcl interface (generated by swig)

Classes

Functions

Function Documentation