cmtk::BestDirectionOptimizer Class Reference [cmtkRegistration Library]

Optimizer derived from BestNeighbourOptimizer. More...

#include <cmtkBestDirectionOptimizer.h>

Inheritance diagram for cmtk::BestDirectionOptimizer:

Collaboration diagram for cmtk::BestDirectionOptimizer:

List of all members.

Public Types
typedef BestDirectionOptimizer	Self
	This class.
typedef Optimizer	Superclass
	Superclass.
Public Member Functions
	cmtkGetSetMacroDefault (bool, UseMaxNorm, true)
	Flag whether to use maximum (1) or Euclid (0) for normalization.
	cmtkGetSetMacroDefault (Self::ParameterType, DirectionThreshold,-1)
	Threshold for direction components.
	cmtkGetSetMacro (int, RepeatLevelCount)
	Number of repetitions of each search level, even if previously unsuccessful.
	cmtkGetSetMacro (bool, AggressiveMode)
	Agressive mode.
	BestDirectionOptimizer (const Self::ParameterType stepFactor=0.5, const Self::ParameterType=0.1)
	Constructor.
virtual CallbackResult	Optimize (CoordinateVector &, const Self::ParameterType=1, const Self::ParameterType=0)
	Optimize functional.
Private Attributes
Self::ParameterType	StepFactor
	Factor by which the step size is reduced after each pass.

---

Detailed Description

Optimizer derived from BestNeighbourOptimizer.

Definition at line 49 of file cmtkBestDirectionOptimizer.h.

---

Member Typedef Documentation

typedef BestDirectionOptimizer cmtk::BestDirectionOptimizer::Self

This class.

Reimplemented from cmtk::Optimizer.

Definition at line 55 of file cmtkBestDirectionOptimizer.h.

typedef Optimizer cmtk::BestDirectionOptimizer::Superclass

Superclass.

Reimplemented from cmtk::Optimizer.

Definition at line 58 of file cmtkBestDirectionOptimizer.h.

---

Constructor & Destructor Documentation

cmtk::BestDirectionOptimizer::BestDirectionOptimizer	(	const Self::ParameterType	stepFactor = 0.5,
		const Self::ParameterType	= 0.1
	)		[inline]

Constructor.

Definition at line 90 of file cmtkBestDirectionOptimizer.h.

---

Member Function Documentation

cmtk::BestDirectionOptimizer::cmtkGetSetMacro	(	int	,
		RepeatLevelCount
	)

Number of repetitions of each search level, even if previously unsuccessful.

This is one by default, so whenever no successful update was made, the level is finished. Setting this to values larger than 1 only makes sense if the optimized functional is changing over time, e.g., due to probabilistic effects.

cmtk::BestDirectionOptimizer::cmtkGetSetMacro	(	bool	,
		AggressiveMode
	)

Agressive mode.

If this flag is set, the optimization is continued at one level as long as there is an improvement to the target function at any step level. Otherwise, steps in the binary seach phase are not considered and search terminates earlier.

cmtk::BestDirectionOptimizer::cmtkGetSetMacroDefault	(	Self::ParameterType	,
		DirectionThreshold	,
		-	1
	)

Threshold for direction components.

Before searching in a certain directions, all components below this fraction of the maximum absolute component are set to zero. This is done to prevent changes in irrelevant parameters and thus support local recomputation when optimizing elastic deformation parameters for example. set this flag to 0 to disable thresholding; set it to 1 to remove all but the most significant components.

cmtk::BestDirectionOptimizer::cmtkGetSetMacroDefault	(	bool	,
		UseMaxNorm	,
		true
	)

Flag whether to use maximum (1) or Euclid (0) for normalization.

---

Member Data Documentation

Self::ParameterType cmtk::BestDirectionOptimizer::StepFactor [private]

Factor by which the step size is reduced after each pass.

Definition at line 104 of file cmtkBestDirectionOptimizer.h.

---

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

• libs/Registration/cmtkBestDirectionOptimizer.h
• libs/Registration/cmtkBestDirectionOptimizer.cxx