jafar::gfm_v2::DescriptorsBasedMatcher< InterestFeatureT > Class Template Reference

This class implements a variation of Jung group matching algorithm. More...

---

Detailed Description

template<typename InterestFeatureT>
class jafar::gfm_v2::DescriptorsBasedMatcher< InterestFeatureT >

This class implements a variation of Jung group matching algorithm.

Definition at line 18 of file DescriptorsBasedMatcher.hpp.

#include <DescriptorsBasedMatcher.hpp>

Inheritance diagram for jafar::gfm_v2::DescriptorsBasedMatcher< InterestFeatureT >:

List of all members.

Classes
class	DescriptorsMatchSourceInfo
struct	matched_feature
	a matched feature consist of an observed feature of match and candidates in ref More...
struct	similar_feature
	a similar feature is a feature from ref lying at distance from match More...
Public Types
typedef InterestFeatureT::DescriptorType::RealType	RealType
typedef DescriptorsBasedMatcher < InterestFeatureT >	MatcherType
typedef fdetect_v2::DetectionResult < InterestFeatureT >	DetectionResult
typedef gfm_v2::MatchingResult < InterestFeatureT >	MatchingResult
typedef gfm_v2::MatchSourceInfo < InterestFeatureT >	MatchSourceInfo
Public Member Functions
virtual MatchSourceInfo *	initMatchSourceInfo (fdetect_v2::DetectionResult< InterestFeatureT > &pointsource)
virtual gfm_v2::MatchingResult < InterestFeatureT >	computeMatch (gfm_v2::MatchSourceInfo< InterestFeatureT > *refsourceinfo, gfm_v2::MatchSourceInfo< InterestFeatureT > *matchsourceinfo)
	Compute the matching between two sources of information.
	DescriptorsBasedMatcher (int knn=2, int checks=20, RealType threshold=0.8, bool enforce=false, RealType increment=1.5, int bins=20)
	Constructor.
RealType	distance (unsigned int matched_index)
	distance between matched and reference feature
Static Public Member Functions
static void	vote_size_orientation (gfm_v2::MatchingResult< InterestFeatureT > &mr, RealType scale_increment=1.5, int rotation_bins=20)
	vote for size and orientation
Private Member Functions
void	brute_force (gfm_v2::MatchSourceInfo< InterestFeatureT > *refsourceinfo, gfm_v2::MatchSourceInfo< InterestFeatureT > *matchsourceinfo, std::vector< matched_feature > &matches, int knn, int checks)
	performs a brute force k nearest neighbours search
std::vector< matched_feature >	vote_oneness (std::vector< matched_feature > &matches, RealType threshold)
	vote for uniqueness of features
Private Attributes
int	m_knn
int	m_checks
RealType	m_uniquess_threshold
bool	m_enforce
RealType	m_increment
int	m_bins
std::map< unsigned int, RealType >	m_distances

---

Constructor & Destructor Documentation

template<typename InterestFeatureT >

jafar::gfm_v2::DescriptorsBasedMatcher< InterestFeatureT >::DescriptorsBasedMatcher	(	int	knn = 2,
		int	checks = 20,
		RealType	threshold = 0.8,
		bool	enforce = false,
		RealType	increment = 1.5,
		int	bins = 20
	)		[inline]

Constructor.

Parameters:

knn	number of neighbours to be sought for knn search algo
checks	how many leafs to check for knn search algo
threshold	threshold to differencaite two close features
enforce	use size and orientation vote to filter result
increment	scale increment used in size and orientation vote
bins	rotation nbr of bins used in size and orientation vote

Definition at line 86 of file DescriptorsBasedMatcher.hpp.

---

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

• modules/gfm/include/gfm/DescriptorsBasedMatcher.hpp