mimas::svd_correspondence< InputIter1, SimilarityMeasure > Class Template Reference
[Tracking of Features]

Correspondence by Singular Value Decomposition. More...

#include <svd_correspondence.h>

Inheritance diagram for mimas::svd_correspondence< InputIter1, SimilarityMeasure >:

[legend]Collaboration diagram for mimas::svd_correspondence< InputIter1, SimilarityMeasure >:


Public Types
typedef boost::numeric::ublas::matrix< double, boost::numeric::ublas::column_major >	Matrix
typedef boost::numeric::ublas::identity_matrix< double >	IdentityMatrix
Public Member Functions
	svd_correspondence (SimilarityMeasure &aSimilarityMeasure, InputIter1 aBegin1, int aSize1)
	Constructor.
template<class InputIter2, class OutputIter1, class OutputIter2>
void	match (InputIter2 begin2, int size2, OutputIter1 outputIndexed, OutputIter2 outputLost) throw (mimasexception)
	Match against new feature set.
Protected Attributes
SimilarityMeasure	similarityMeasure
	Similarity measure.
InputIter1	begin1
	Iterator on indexed feature set.
int	size1
	Size of indexed feature set.

Detailed Description

template<class InputIter1, class SimilarityMeasure>
class mimas::svd_correspondence< InputIter1, SimilarityMeasure >

Correspondence by Singular Value Decomposition.

This class implements the core algorithm described in Maurizio Pilu's article.

Given

a map $m:I\mapsto F$ where is an index set and is a set of features
a second set of features $F^\prime$ .
a similarity measure $F\times F^\prime\mapsto [0,1]$ giving real numbers between zero and one

the algorithm will guess about correspondences and yield a new map $m^\prime:I^\prime\mapsto F^\prime$ , where $I^\prime\subset I$ . If the algorithm detects ambiguity, it will drop some indices and $I^\prime$ will be a true subset of $I$

InputIter1 must implement the concept of Input Iterator and it must be an iterator on a pair with an index as first and a feature as second element. Usually this will be something like

    map< Index, F >::const_iterator

where F denotes the feature class.

SimilarityMeasure must be a binary_function< F, F, double >, where F denotes the feature class.

Author:: Jan Wedekind (jan at wedesoft.de)

Date:: Thu May 20 15:53:37 UTC 2004

Definition at line 47 of file svd_correspondence.h.

Member Typedef Documentation

template<class InputIter1, class SimilarityMeasure>

typedef boost::numeric::ublas::matrix< double, boost::numeric::ublas::column_major > mimas::svd_correspondence< InputIter1, SimilarityMeasure >::Matrix

Definition at line 51 of file svd_correspondence.h.

template<class InputIter1, class SimilarityMeasure>

typedef boost::numeric::ublas::identity_matrix< double > mimas::svd_correspondence< InputIter1, SimilarityMeasure >::IdentityMatrix

Definition at line 52 of file svd_correspondence.h.

Constructor & Destructor Documentation

template<class InputIter1, class SimilarityMeasure>

mimas::svd_correspondence< InputIter1, SimilarityMeasure >::svd_correspondence	(	SimilarityMeasure &	aSimilarityMeasure,
		InputIter1	aBegin1,
		int	aSize1
	)			`[inline]`

Constructor.

The constructor will take the index-map and the similarity measure.

Parameters:

	aSimilarityMeasure	Similarity measure, giving values $\in[0,1]$ .
	aBegin1	Iterator on feature-map .
	aSize1	Number of elements in .

Definition at line 59 of file svd_correspondence.h.

Member Function Documentation

template<class InputIter1, class SimilarityMeasure>

template<class InputIter2, class OutputIter1, class OutputIter2>

void mimas::svd_correspondence< InputIter1, SimilarityMeasure >::match	(	InputIter2	begin2,
		int	size2,
		OutputIter1	outputIndexed,
		OutputIter2	outputLost
	)			throw (mimasexception)

Match against new feature set.

This method will take a feature set $F^\prime$ containing new features. The method will give the result by writing index-feature-pairs to output.

Parameters:

	begin2	Input iterator on new feature set.
	size2	Size of new feature set.
	outputIndexed	Output iterator to store indexed features in (usually an object of type `map<>::iterator`).
	outputLost	Output iterator to store lost features in.

Member Data Documentation

template<class InputIter1, class SimilarityMeasure>

SimilarityMeasure mimas::svd_correspondence< InputIter1, SimilarityMeasure >::similarityMeasure [protected]

Similarity measure.

An object of type binary_function<> giving similarity estimates as double values.

Definition at line 78 of file svd_correspondence.h.

template<class InputIter1, class SimilarityMeasure>

InputIter1 mimas::svd_correspondence< InputIter1, SimilarityMeasure >::begin1 [protected]

Iterator on indexed feature set.

Definition at line 80 of file svd_correspondence.h.

template<class InputIter1, class SimilarityMeasure>

int mimas::svd_correspondence< InputIter1, SimilarityMeasure >::size1 [protected]

Size of indexed feature set.

Definition at line 82 of file svd_correspondence.h.

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

svd_correspondence.h

mimas 2.1

Copyright Mon Oct 30 11:31:48 2006, Bala Amavasai, Stuart Meikle, Arul Selvan, Fabio Caparrelli, Jan Wedekind, Manuel Boissenin, ...

mimas::svd_correspondence< InputIter1, SimilarityMeasure > Class Template Reference [Tracking of Features]

Public Types

Public Member Functions

Protected Attributes

Detailed Description

template<class InputIter1, class SimilarityMeasure> class mimas::svd_correspondence< InputIter1, SimilarityMeasure >

Member Typedef Documentation

Constructor & Destructor Documentation

Member Function Documentation

Member Data Documentation

mimas::svd_correspondence< InputIter1, SimilarityMeasure > Class Template Reference
[Tracking of Features]

template<class InputIter1, class SimilarityMeasure>
class mimas::svd_correspondence< InputIter1, SimilarityMeasure >