lti::gaussianPyramid< T > Class Template Reference

GaussianPyramid class. More...

#include <ltiGaussianPyramid.h>

Inheritance diagram for lti::gaussianPyramid< T >:

[legend]

Collaboration diagram for lti::gaussianPyramid< T >:

[legend]

List of all members.

Public Member Functions
	gaussianPyramid (const int &resolutions=0, const int &gaussianSize=3, const double &variance=-1, const bool &upsampleWithGaussian=true)
	gaussianPyramid (const gaussianPyramid &other)
virtual	~gaussianPyramid ()
void	getKernelParameters (int &size, double &variance, bool &gaussian) const
void	setKernelParameters (const int &size, const double &variance=-1, const bool &gaussian=true)
const char *	getTypeName () const
gaussianPyramid< T > &	copy (const gaussianPyramid< T > &other)
virtual mathObject *	clone () const
void	generate (const T &src)
void	generate (const T &src, const int &theResolutions)
void	generate (const T &src, const lti::point &limit)
T &	reconstruct (const int &i, const int &fromJ, T &result) const
pyramid< T > &	reconstruct (const int &i, const int &fromJ, pyramid< T > &result) const
Protected Attributes
int	kernelSize
double	kernelVariance
bool	gaussian

Detailed Description

template<class T>
class lti::gaussianPyramid< T >

GaussianPyramid class.

This class implements the gaussian pyramids as described in Greenspan et.at. "Overcomplete Steerable Pyramid Filters and Rotation Invariance", Proc. of the IEEE Conf. on Computer Visiona and Pattern Recognition, Seattle, 1994

It allows an easy manipulation of the different resolutions.

See method generate() to an detailed explanation.

The template type T is the type of the elements in the pyramid.

Example:

 channel chnl;
 // initialize channel or load it from an image...

 lti::gaussianPyramid<lti::channel> thePyramid(4); // four levels
 thePyramid.generate(chnl); // generate a pyramid for the given channel

Constructor & Destructor Documentation

template<class T>

lti::gaussianPyramid< T >::gaussianPyramid	(	const int &	resolutions = `0`,
		const int &	gaussianSize = `3`,
		const double &	variance = `-1`,
		const bool &	upsampleWithGaussian = `true`
	)

create a gaussian pyramid with the given number of resolutions

Parameters:

	resolutions	the number of resolutions that the pyramid can hold (default 0: an empty pyramid will be created)
	upsampleWithGaussian	a gaussian kernel will be used to upsample the images. If false a squared kernel will be used.
	gaussianSize	the size of the gaussian kernel (default 3)
	variance	the variance of the gaussian kernel (default -1, meaning that the variance should be calculated as described in the lti::gaussKernel1D<T>)

template<class T>

lti::gaussianPyramid< T >::gaussianPyramid ( const gaussianPyramid< T > & other )

create this gaussianPyramid as a copy of another gaussianPyramid

Parameters:

other

the gaussianPyramid to be copied.

template<class T>

virtual lti::gaussianPyramid< T >::~gaussianPyramid ( ) [virtual]

destructor

Member Function Documentation

template<class T>

virtual mathObject* lti::gaussianPyramid< T >::clone ( ) const [virtual]

create a clone of this gaussianPyramid

Returns:: a pointer to a copy of this gaussianPyramid

Reimplemented from lti::pyramid< T >.

Reimplemented in lti::gaborPyramid< T >.

template<class T>

gaussianPyramid<T>& lti::gaussianPyramid< T >::copy ( const gaussianPyramid< T > & other )

assigment operator.

copy the contents of other in this object.

Parameters:

other

the source gaussianPyramid to be copied.

Returns:: a reference to this object

Reimplemented from lti::pyramid< T >.

Reimplemented in lti::gaborPyramid< T >.

template<class T>

void lti::gaussianPyramid< T >::generate	(	const T &	src,
		const lti::point &	limit
	)

generate the gaussian pyramid of the given object.

Proceed, until given limit.x or limit.y is reached. smallest resolution will be > limit.

The resolution "0" corresponds to the original channel, and the resolution i+1 is always a factor 2 smaller than the resolution i.

Reimplemented in lti::gaborPyramid< T >.

template<class T>

void lti::gaussianPyramid< T >::generate	(	const T &	src,
		const int &	theResolutions
	)

generate the gaussian pyramid of the given object.

The pyramid will contain the number of resolutions specified by theResolutions. The resolution "0" will correspond the the original channel, and the resolution i+1 is always a factor 2 smaller than the resolution i.

Reimplemented in lti::gaborPyramid< T >.

template<class T>

void lti::gaussianPyramid< T >::generate ( const T & src )

generate the gaussian pyramid of the given object.

The pyramid will contain the number of resolutions specified in the construction or in the resize() method. The resolution "0" will correspond the the original channel, and the resolution i+1 is always a factor 2 smaller than the resolution i.

Reimplemented in lti::gaborPyramid< T >.

template<class T>

void lti::gaussianPyramid< T >::getKernelParameters	(	int &	size,
		double &	variance,
		bool &	gaussian
	)			const

return the size and variance of the used gaussian kernel

template<class T>

const char* lti::gaussianPyramid< T >::getTypeName ( void ) const [inline, virtual]

returns the name of this class: "gaussianPyramid"

Reimplemented from lti::pyramid< T >.

Reimplemented in lti::gaborPyramid< T >.

template<class T>

pyramid<T>& lti::gaussianPyramid< T >::reconstruct	(	const int &	i,
		const int &	fromJ,
		pyramid< T > &	result
	)			const

reconstructs the resolution with index i from a second resolution j.

If i>j, the object the same pyramid will be returned, otherwise the corresponding upsampling will be done with the kernel type specified in the construction A pyramid with the same original size will be returned, where only the elements between i and fromJ will be initialized.

Parameters:

	i	the resolution to be reconstructed
	fromJ	the resolution from which the data is to be taken.
	result	the resulting pyramid.

Returns:: a reference to the result pyramid

template<class T>

T& lti::gaussianPyramid< T >::reconstruct	(	const int &	i,
		const int &	fromJ,
		T &	result
	)			const

reconstructs the resolution with index i from a second resolution j.

If i>j, the object at(i) will be returned, otherwise the corresponding upsampling will be done with the kernel type specified in the construction

Parameters:

	i	the resolution to be reconstructed
	fromJ	the resolution from which the data is to be taken.
	result	the resulting object

Returns:: a reference to the result object.

template<class T>

void lti::gaussianPyramid< T >::setKernelParameters	(	const int &	size,
		const double &	variance = `-1`,
		const bool &	gaussian = `true`
	)

set the kernel parameters

Parameters:

	size	the size of the kernel
	variance	the variance for the gaussian kernel. A negative value will force the default variance of a gaussian kernel with size `size`.
	gaussian	specify if for the channel upsampling a gaussian or a rectangular kernel should be used.