drwnNNGraphThreadedMoves Namespace Reference

Templated drwnNNGraph search moves (multi-threaded). More...

Classes
class	drwnNNGraphThreadedInitializeJob
	threading initialize functor More...
class	drwnNNGraphThreadedRescoreJob
	threading rescore functor More...
class	drwnNNGraphThreadedUpdateJob
	threading update functor More...
class	drwnNNGraphThreadedExhaustiveJob
	threading exhaustive functor More...

Functions
template<class DistanceMetric >
void	initialize (drwnNNGraph &graph, const DistanceMetric &M)
	randomly initialize edges (matched) for all active images (keeps existing matches unless an improvement is found)
void	initialize (drwnNNGraph &graph)
	randomly initialize edges (matched) for all active images (keeps existing matches unless an improvement is found) using default metric
template<class DistanceMetric >
double	rescore (drwnNNGraph &graph, const DistanceMetric &M)
	rescore all matches based on current features (which may have changed since graph was created)
double	rescore (drwnNNGraph &graph)
	rescore all matches based on current features (which may have changed since graph was created) using default metric
template<class DistanceMetric >
void	update (drwnNNGraph &graph, const DistanceMetric &M)
	perform one update iteration (including enrichment and exhaustive) on all active images
void	update (drwnNNGraph &graph)
	perform one update iteration (including enrichment and exhaustive) on all active images using default metric
template<class DistanceMetric >
void	updateImage (drwnNNGraph &graph, unsigned imgIndx, const DistanceMetric &M)
	perform single update iteration of propagate, local and search moves on a specific image
template<class DistanceMetric >
bool	randproj (drwnNNGraph &graph, const DistanceMetric &M)
	random projection moves that project all superpixels onto a random direction, sort and compare adjacent superpixels if from different images (returns true if a better match was found)
template<class DistanceMetric >
bool	enrichment (drwnNNGraph &graph, const DistanceMetric &M)
	enrichment: inverse (from target to source) and forward (from source to target's target) (returns true if a better match was found)

Detailed Description

Templated drwnNNGraph search moves (multi-threaded).