Graph Extensions

A set of wrappers for building graphical models for 2D image classifaction. More...

Collaboration diagram for Graph Extensions:

Classes
class	DirectGraphicalModels::CGraphDenseExt
	Extended Dense graph class for 2D image classifaction. More...
class	DirectGraphicalModels::CGraphExt
	General graph extension abstract class for 2D image classifaction. More...
class	DirectGraphicalModels::CGraphLayeredExt
	Extended Pairwise Layered graph class. More...
class	DirectGraphicalModels::CGraphPairwiseExt
	Extended Pairwise graph class for 2D image classifaction. More...

Detailed Description

A set of wrappers for building graphical models for 2D image classifaction.

For simplifying the application of the DGM library to the classification of the images, this sub-module provides a set of additional classes which extend the functionality of the Graph Building module. A graph representing an image usually has 2D grid structure with the standard edge pattern (in case of pairwise graphs). The base class DirectGraphicalModels::CGraphExt provides a set of wrappers for building and filling with potentials such graphs. It also has wrappers for initializing default edge interaction models, which are training-data-independent. This is enough for solving a large variety of pattern recognition and machine vision problems. For adding training-data-dependent edge models one can use the derived from DirectGraphicalModels::CGraphExt classes directly.