lti::curvatureScaleSpace Class Reference

Curvature Scale Space Representation (CSS). More...

#include <ltiCurvatureScaleSpace.h>

Inheritance diagram for lti::curvatureScaleSpace:

[legend]

Collaboration diagram for lti::curvatureScaleSpace:

[legend]

List of all members.

Classes
class	parameters
	the parameters for the class curvatureScaleSpace More...
Public Types
enum	eFeatureIndices { MaxY = 0, MaxX, Width, Offset }
Public Member Functions
	curvatureScaleSpace ()
	curvatureScaleSpace (const curvatureScaleSpace &other)
virtual	~curvatureScaleSpace ()
virtual const char *	getTypeName () const
bool	apply (const borderPoints &src, channel8 &dest, float &maxSigma, int &maxRow) const
bool	apply (const borderPoints &src, channel8 &dest, pointList &rows) const
bool	apply (const borderPoints &src, channel8 &dest) const
bool	apply (const borderPoints &src, channel8 &css, dmatrix &vcts) const
bool	apply (const borderPoints &src, channel8 &css, dmatrix &vcts, pointList &rows) const
bool	apply (const borderPoints &src, dmatrix &vcts) const
curvatureScaleSpace &	copy (const curvatureScaleSpace &other)
virtual functor *	clone () const
const parameters &	getParameters () const
Protected Member Functions
bool	borderToXY (const borderPoints &src, vector< float > &theX, vector< float > &theY) const
void	calcKernels (const float &sigma, kernel1D< float > &first, kernel1D< float > &second) const
int	curvature (const float &sigma, const vector< float > &theX, const vector< float > &theY, vector< float > &curv, vector< ubyte > &zeroCross) const
void	calcKernels (float &sigma, gaussKernel1D< float > &gaussian) const
int	curvature (const gaussKernel1D< float > &gaussian, vector< float > &theX, vector< float > &theY, vector< float > &curv, vector< ubyte > &zeroCross, const int &lastZC) const
bool	cssOriginal (const borderPoints &src, channel8 &dest, float &maxSigma, int &maxRow, pointList &rows) const
bool	cssFast (const borderPoints &src, channel8 &dest, float &maxSigma, int &maxRow, pointList &rows) const
int	zeroCross (const vector< float > &curv, vector< ubyte > &dest, const ubyte &posNeg, const ubyte &negPos) const
int	extractBlobs (const channel8 &css, const pointList &rows, matrix< int > &blobs, dmatrix &features) const

Detailed Description

Curvature Scale Space Representation (CSS).

This functor extracts from a boundary (borderPoints) the CSS representation as described in Mokhtarian, F. et. al "Robust and efficient shape indexing through curvature scale space", British Machine Vision Conference, 1996

The apply methods will check if there are enough points in the input parameters to compute the css. If there are not enough points, nothing will be done (see the apply methods for more details).

Member Enumeration Documentation

enum lti::curvatureScaleSpace::eFeatureIndices

indices for the columns of the feature-vectors in the matrix

Enumerator:

MaxY	maximum sigma value of the blob
MaxX	position x (length index) of the blob maximum
Width	average width of the blob
Offset	diference between the average width position and the position x of the maximum

Constructor & Destructor Documentation

lti::curvatureScaleSpace::curvatureScaleSpace ( )

default constructor

lti::curvatureScaleSpace::curvatureScaleSpace ( const curvatureScaleSpace & other )

copy constructor

Parameters:

other

the object to be copied

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

destructor

Member Function Documentation

bool lti::curvatureScaleSpace::apply	(	const borderPoints &	src,
		dmatrix &	vcts
	)			const

operates on the given parameter.

This member computes the CSS-image and the feature vectors. Each row of the matrix will correspond to a feature vector which contains:

Column 0: (alias for 0: MaxY) the value of the maximum of the blob.
Column 1: (alias for 1: MaxX) the x-position of maximum of the blob.
Column 2: (alias for 2: Width) the average width
Column 3: (alias for 3: Offset) the offset between the x-position of the maximum of the blob and the center of the average width.

Parameters:

	src	list of boundary points. If this list has less than 9 points (the smallest "circle" has 8 points), false will be returned with the respective message in the status string
	vcts	feature vectors. Each row corresponds to a CSS "blob"

Returns:: true if apply successful or false otherwise.

bool lti::curvatureScaleSpace::apply	(	const borderPoints &	src,
		channel8 &	css,
		dmatrix &	vcts,
		pointList &	rows
	)			const

operates on the given parameter.

This member computes the CSS-image and the feature vectors. Each row of the matrix will correspond to a feature vector which contains:

Column 0: (alias for 0: MaxY) the value of the maximum of the blob.
Column 1: (alias for 1: MaxX) the x-position of maximum of the blob.
Column 2: (alias for 2: Width) the average width
Column 3: (alias for 3: Offset) the offset between the x-position of the maximum of the blob and the center of the average width.

Parameters:

	src	list of boundary points. If this list has less than 9 points (the smallest "circle" has 8 points), false will be returned with the respective message in the status string
	css	the CSS image will be left here.
	vcts	the feature vectors packed in a matrix.
	rows	a list of points p, where p.x contains the number of blobs crossing the row of dest with index p.y. In a "clean" CSS representation the number of blobs has to decrease or be equal with increasing index p.y. If you detect here the opposite effect is because there is too much noise in the high-resolution levels (small variance of Gaussians).

Returns:: true if apply successful or false otherwise.

bool lti::curvatureScaleSpace::apply	(	const borderPoints &	src,
		channel8 &	css,
		dmatrix &	vcts
	)			const

operates on the given parameter.

This member computes the CSS-image and the feature vectors. Each row of the matrix will correspond to a feature vector which contains:

Column 0: (alias for 0: MaxY) the value of the maximum of the blob.
Column 1: (alias for 1: MaxX) the x-position of maximum of the blob.
Column 2: (alias for 2: Width) the average width
Column 3: (alias for 3: Offset) the offset between the x-position of the maximum of the blob and the center of the average width.

Parameters:

	src	list of boundary points. If this list has less than 9 points (the smallest "circle" has 8 points), false will be returned with the respective message in the status string
	css	the CSS image will be left here.
	vcts	the feature vectors packed in a matrix.

Returns:: true if apply successful or false otherwise.

bool lti::curvatureScaleSpace::apply	(	const borderPoints &	src,
		channel8 &	dest
	)			const

operates on the given parameter.

Parameters:

	src	list of boundary points. If this list has less than 9 points (the smallest "circle" has 8 points), false will be returned with the respective message in the status string
	dest	the CSS image will be left here.

Returns:: true if apply successful or false otherwise.

bool lti::curvatureScaleSpace::apply	(	const borderPoints &	src,
		channel8 &	dest,
		pointList &	rows
	)			const

operates on the given parameter.

Parameters:

	src	list of boundary points. If this list has less than 9 points (the smallest "circle" has 8 points), false will be returned with the respective message in the status string
	dest	the CSS image will be left here.
	rows	a list of points p, where p.x contains the number of blobs crossing the row of dest with index p.y. In a "clean" CSS representation the number of blobs has to decrease or be equal with increasing index p.y. If you detect here the opposite effect is because there is too much noise in the high-resolution levels (small variance of Gaussians).

Returns:: true if apply successful or false otherwise.

bool lti::curvatureScaleSpace::apply	(	const borderPoints &	src,
		channel8 &	dest,
		float &	maxSigma,
		int &	maxRow
	)			const

operates on the given parameter.

Parameters:

	src	list of boundary points. If this list has less than 9 points (the smallest "circle" has 8 points), false will be returned with the respective message in the status string
	dest	the CSS image will be left here.
	maxSigma	the maximum sigma in the CSS image.
	maxRow	the row corresponding to the maximum sigma.

Returns:: true if apply successful or false otherwise.

bool lti::curvatureScaleSpace::borderToXY	(	const borderPoints &	src,
		vector< float > &	theX,
		vector< float > &	theY
	)			const `[protected]`

extract the x and y components of the borderPoints object and resize (normalize the length) of the vectors if required.

void lti::curvatureScaleSpace::calcKernels	(	float &	sigma,
		gaussKernel1D< float > &	gaussian
	)			const `[protected]`

calculate the gaussian first and second derivative kernels.

The size of the kernel is recomputed considering the parameters object. The sigma term will be updated to the new one.

void lti::curvatureScaleSpace::calcKernels	(	const float &	sigma,
		kernel1D< float > &	first,
		kernel1D< float > &	second
	)			const `[protected]`

calculate the gaussian first and second derivative kernels.

The size is calcultated considering the parameters object.

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

returns a pointer to a clone of this functor.

Implements lti::functor.

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

copy data of "other" functor.

Parameters:

other

the functor to be copied

Returns:: a reference to this functor object

Reimplemented from lti::functor.

bool lti::curvatureScaleSpace::cssFast	(	const borderPoints &	src,
		channel8 &	dest,
		float &	maxSigma,
		int &	maxRow,
		pointList &	rows
	)			const `[protected]`

operates on the given parameter.

Parameters:

	src	list of boundary points.
	dest	modified CSS image will be left here.
	maxSigma	the maximum sigma in the CSS image.
	maxRow	the row corresponding to the maximum sigma.
	rows	a list of points, where x contains the number of maxima of the row with the index given in the point's y.

Returns:: true if apply successful or false otherwise.

bool lti::curvatureScaleSpace::cssOriginal	(	const borderPoints &	src,
		channel8 &	dest,
		float &	maxSigma,
		int &	maxRow,
		pointList &	rows
	)			const `[protected]`

operates on the given parameter.

Parameters:

	src	list of boundary points.
	dest	the CSS image will be left here.
	maxSigma	the maximum sigma in the CSS image.
	maxRow	the row corresponding to the maximum sigma.
	rows	a list of points, where x contains the number of maxima of the row with the index given by y.

Returns:: true if apply successful or false otherwise.

int lti::curvatureScaleSpace::curvature	(	const gaussKernel1D< float > &	gaussian,
		vector< float > &	theX,
		vector< float > &	theY,
		vector< float > &	curv,
		vector< ubyte > &	zeroCross,
		const int &	lastZC
	)			const `[protected]`

calculate the curvature using the derivatives

int lti::curvatureScaleSpace::curvature	(	const float &	sigma,
		const vector< float > &	theX,
		const vector< float > &	theY,
		vector< float > &	curv,
		vector< ubyte > &	zeroCross
	)			const `[protected]`

calculate the curvature using the derivatives

int lti::curvatureScaleSpace::extractBlobs	(	const channel8 &	css,
		const pointList &	rows,
		matrix< int > &	blobs,
		dmatrix &	features
	)			const `[protected]`

extract the blobs from the css image (only the first 127 blobs will be extracted)

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

returns used parameters

Reimplemented from lti::transform.

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

returns the name of this type ("curvatureScaleSpace")

Reimplemented from lti::transform.

int lti::curvatureScaleSpace::zeroCross	(	const vector< float > &	curv,
		vector< ubyte > &	dest,
		const ubyte &	posNeg,
		const ubyte &	negPos
	)			const `[protected]`

calculate the zero crossings.

The output vector MUST have the same size as the input vector.

This member returns the number of zero crossings detected.

The documentation for this class was generated from the following file:

ltiCurvatureScaleSpace.h

lti::curvatureScaleSpace Class Reference

Classes

Public Types

Public Member Functions

Protected Member Functions

Detailed Description

Member Enumeration Documentation

Constructor & Destructor Documentation

Member Function Documentation