cvr::fastHessianDetection Class Reference
[Feature Extraction]

#include <cvrFastHessianDetection.h>

Inheritance diagram for cvr::fastHessianDetection:

Collaboration diagram for cvr::fastHessianDetection:

List of all members.

Classes

class parameters

The parameters for the class fastHessianDetection. More...

Public Types

enum eExtremaType { Minima, Maxima, Both }

enum eLevelSelectionMethod { Blocks, Exponential }

enum eLocationSelectionMode {
All, Absolute, Relative, Conspicuous,
Number
}

Public Member Functions

fastHessianDetection ()

fastHessianDetection (const parameters &par)

fastHessianDetection (const fastHessianDetection &other)

virtual ~fastHessianDetection ()

virtual bool apply (const channel8 &src, list< location > &locs) const

bool apply (const channel8 &src, list< location > &locs, list< float > &strength) const

bool apply (const channel8 &src, list< location > &locs, list< float > &strength, int &numLocs) const

virtual bool apply (const channel8 &src, list< location > &locs, int &numLocs) const

virtual bool apply (const channel &src, list< location > &locs) const

bool apply (const channel &src, list< location > &locs, list< float > &strength) const

bool apply (const channel &src, list< location > &locs, list< float > &strength, int &numLocs) const

virtual bool apply (const channel &src, list< location > &locs, int &numLocs) const

bool sort (const list< location > &locs, const list< float > &strength, std::vector< location > &sortedLocs, fvector &sortedStrength) const

fastHessianDetection & copy (const fastHessianDetection &other)

fastHessianDetection & operator= (const fastHessianDetection &other)

virtual const std::string & name () const

virtual fastHessianDetection * clone () const

virtual fastHessianDetection * newInstance () const

const parameters & getParameters () const

bool updateParameters ()

Detailed Description

Class fastHessianDetection.

The class cvr::fastHessianDetection is one of the available location detectors in the CVR-Lib , i.e. is a class used to detect interesing points in an image, considering also the "scale" at which the interesting information resides.

This particular method is part of the SURF approach for robust feature detection, as explained in detail in the paper:

Herbert Bay, Tinne Tuytelaars, and Luc Van Gool. SURF: Speeded Up Robust Features. In Proceedings of the 9th European Conference on Computer Vision, May 2006. Available at http://www.vision.ee.ethz.ch/~surf/papers.html

The implementation here provides some configuration possibilities, but is limited to provide the authors concept as reliable as the paper allowed.

See also:: fastHessianDetection::parameters.
surfLocalDescriptor

Member Enumeration Documentation

enum cvr::fastHessianDetection::eExtremaType

Types for locations to be found.

Enumerator:

Minima	Only detect minima of the determinant image.
Maxima	Only detect maxima of the determinant image.
Both	Detect both maxima and minima.

enum cvr::fastHessianDetection::eLevelSelectionMethod

Types of level selection in the multiresolutional representation.

Enumerator:

Blocks	Original method suggested in the SURF paper, which uses blocks of levels separated by the same scale step, After a block has been computed, the next one duplicates the step.
Exponential	The exponential level selection method uses for the scales the equation $\alpha^n\sigma_B$ .

enum cvr::fastHessianDetection::eLocationSelectionMode

Modes available for the selection of conspicuous locations.

These modes affect the way in which the threshold value is interpreted.

Enumerator:

All	Ignore the threshold, and compute all locations.
Absolute	The determinant of the hessian matrix has to be greater than the threshold value (or less than -threshold) to be considered.
Relative	The determinant of the hessian matrix has to be greater than a threshold value computed as a percentage of the greatest value found at each level.
Conspicuous	The determinant of the hessian matrix has to be greater than a threshold value computed as the mean of all values plus a number of times the standard deviation.
Number	At most the given number of (strongest) locations will be returned.

Constructor & Destructor Documentation

cvr::fastHessianDetection::fastHessianDetection ( )

Default constructor.

cvr::fastHessianDetection::fastHessianDetection ( const parameters & par )

Construct a functor using the given parameters.

cvr::fastHessianDetection::fastHessianDetection ( const fastHessianDetection & other )

Copy constructor.

Parameters:

other

the object to be copied

virtual cvr::fastHessianDetection::~fastHessianDetection ( ) [virtual]

Destructor.

Member Function Documentation

virtual bool cvr::fastHessianDetection::apply	(	const channel &	src,
		list< location > &	locs,
		int &	numLocs
	)			const `[virtual]`

Compute the locations based on the determinant of the "fast Hessian" matrix.

This method provides numLocs = locs.size() directly. The reason is simple: speed. The size() method needs to count the elements, and this is usually done in the process.

Parameters:

	src	channel8 with the source image.
	locs	lists of detected locations.
	numLocs	number of locations detected.

Returns:: true if apply successful or false otherwise.

Implements cvr::locationDetection.

bool cvr::fastHessianDetection::apply	(	const channel &	src,
		list< location > &	locs,
		list< float > &	strength,
		int &	numLocs
	)			const

Compute the locations based on the determinant of the "fast Hessian" matrix.

Parameters:

	src	channel8 with the source image.
	locs	lists of detected locations.
	strength	the determinant value of the hessian matrix at the center of the location.
	numLocs	number of location in locs and strength

Returns:: true if apply successful or false otherwise.

bool cvr::fastHessianDetection::apply	(	const channel &	src,
		list< location > &	locs,
		list< float > &	strength
	)			const

Compute the locations based on the determinant of the "fast Hessian" matrix.

Parameters:

	src	channel8 with the source image.
	locs	lists of detected locations.
	strength	the determinant value of the hessian matrix at the center of the location.

Returns:: true if apply successful or false otherwise.

virtual bool cvr::fastHessianDetection::apply	(	const channel &	src,
		list< location > &	locs
	)			const `[virtual]`

Compute the locations based on the determinant of the "fast Hessian" matrix.

Parameters:

	src	channel8 with the source image.
	locs	lists of detected locations.

Returns:: true if apply successful or false otherwise.

Implements cvr::locationDetection.

virtual bool cvr::fastHessianDetection::apply	(	const channel8 &	src,
		list< location > &	locs,
		int &	numLocs
	)			const `[virtual]`

Compute the locations based on the determinant of the "fast Hessian" matrix.

This method provides numLocs = locs.size() directly. The reason is simple: speed. The size() method needs to count the elements, and this is usually done in the process.

Parameters:

	src	channel8 with the source image.
	locs	lists of detected locations.
	numLocs	number of locations detected.

Returns:: true if apply successful or false otherwise.

Implements cvr::locationDetection.

bool cvr::fastHessianDetection::apply	(	const channel8 &	src,
		list< location > &	locs,
		list< float > &	strength,
		int &	numLocs
	)			const

Compute the locations based on the determinant of the "fast Hessian" matrix.

Parameters:

	src	channel8 with the source image.
	locs	lists of detected locations.
	strength	the determinant value of the hessian matrix at the center of the location.
	numLocs	number of location in locs and strength

Returns:: true if apply successful or false otherwise.

bool cvr::fastHessianDetection::apply	(	const channel8 &	src,
		list< location > &	locs,
		list< float > &	strength
	)			const

Compute the locations based on the determinant of the "fast Hessian" matrix.

Parameters:

	src	channel8 with the source image.
	locs	lists of detected locations.
	strength	the determinant value of the hessian matrix at the center of the location.

Returns:: true if apply successful or false otherwise.

virtual bool cvr::fastHessianDetection::apply	(	const channel8 &	src,
		list< location > &	locs
	)			const `[virtual]`

Compute the locations based on the determinant of the "fast Hessian" matrix.

Parameters:

	src	channel8 with the source image.
	locs	lists of detected locations.

Returns:: true if apply successful or false otherwise.

Implements cvr::locationDetection.

virtual fastHessianDetection* cvr::fastHessianDetection::clone ( ) const [virtual]

Returns a pointer to a clone of this functor.

Implements cvr::locationDetection.

fastHessianDetection& cvr::fastHessianDetection::copy ( const fastHessianDetection & other )

Operates on a copy of the given arguments.

Parameters:

	src	channel with the source image.
	locs	lists of detected locations.

Returns:: true if apply successful or false otherwise. Copy data of "other" functor.

Parameters:

other

the functor to be copied

Returns:: a reference to this functor object

const parameters& cvr::fastHessianDetection::getParameters ( ) const

Returns used parameters.

Reimplemented from cvr::locationDetection.

virtual const std::string& cvr::fastHessianDetection::name ( ) const [virtual]

Returns the complete name of the functor class.

Implements cvr::locationDetection.

virtual fastHessianDetection* cvr::fastHessianDetection::newInstance ( ) const [virtual]

Returns a pointer to a new instance of this functor.

Implements cvr::locationDetection.

fastHessianDetection& cvr::fastHessianDetection::operator= ( const fastHessianDetection & other )

Alias for copy member.

Parameters:

other

the functor to be copied

Returns:: a reference to this functor object

bool cvr::fastHessianDetection::sort	(	const list< location > &	locs,
		const list< float > &	strength,
		std::vector< location > &	sortedLocs,
		fvector &	sortedStrength
	)			const

Sort the locations according to their strength, in descending order: the strongest locations are always first.

This is useful to take a fixed number of the strongest locations.

The behaviour is influenced by the parameter settings of parameters::extrema. If it is set to Maxima, then the values are sorted as given, in descending order; if set to Minima, then the values are sorted in ascending order; if set to Both, then the absolute value of the strength is sorted in descending order.

Parameters:

	locs	the unsorted locations as provided by the apply method.
	strength	the corresponding strengths for the locations, as provided by the apply method. It has to have exactly the same size as the locs list.
	sortedLocs	the sorted locations as vector, to have access to the rank of the location
	sortedStrength	the sorted strength.

Returns:: true if successful or false otherwise.

bool cvr::fastHessianDetection::updateParameters ( ) [virtual]

Update parameters.

Reimplemented from cvr::parametersManager.

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

cvrFastHessianDetection.h


Classes
class	parameters
	The parameters for the class fastHessianDetection. More...
Public Types
enum	eExtremaType { Minima, Maxima, Both }
enum	eLevelSelectionMethod { Blocks, Exponential }
enum	eLocationSelectionMode { All, Absolute, Relative, Conspicuous, Number }
Public Member Functions
	fastHessianDetection ()
	fastHessianDetection (const parameters &par)
	fastHessianDetection (const fastHessianDetection &other)
virtual	~fastHessianDetection ()
virtual bool	apply (const channel8 &src, list< location > &locs) const
bool	apply (const channel8 &src, list< location > &locs, list< float > &strength) const
bool	apply (const channel8 &src, list< location > &locs, list< float > &strength, int &numLocs) const
virtual bool	apply (const channel8 &src, list< location > &locs, int &numLocs) const
virtual bool	apply (const channel &src, list< location > &locs) const
bool	apply (const channel &src, list< location > &locs, list< float > &strength) const
bool	apply (const channel &src, list< location > &locs, list< float > &strength, int &numLocs) const
virtual bool	apply (const channel &src, list< location > &locs, int &numLocs) const
bool	sort (const list< location > &locs, const list< float > &strength, std::vector< location > &sortedLocs, fvector &sortedStrength) const
fastHessianDetection &	copy (const fastHessianDetection &other)
fastHessianDetection &	operator= (const fastHessianDetection &other)
virtual const std::string &	name () const
virtual fastHessianDetection *	clone () const
virtual fastHessianDetection *	newInstance () const
const parameters &	getParameters () const
bool	updateParameters ()

cvr::fastHessianDetection Class Reference [Feature Extraction]

Classes

Public Types

Public Member Functions

Detailed Description

Member Enumeration Documentation

Constructor & Destructor Documentation

Member Function Documentation

cvr::fastHessianDetection Class Reference
[Feature Extraction]