zxhOptimizerGradient Class Reference

optimizers search for the optimal value of the metric within the transformation follow gradient direction More...

#include <zxhOptimizerGradient.h>

Inheritance diagram for zxhOptimizerGradient:

List of all members.

Public Member Functions
	zxhOptimizerGradient (void)
	constructor
	~zxhOptimizerGradient (void)
	deconstructor
virtual void	SetTransform (zxhTransformBase *p)
	will new a gradient type using transformation type
virtual void	SetGradientClassUsingObjectType (zxhTransformBase *pClassObject)
	set gradient type
virtual std::string	GetPrintString ()
virtual bool	SetOptimizerFromStream (std::ifstream &ifs)
	from stream
virtual zxhOptimizerBase *	Clone (zxhOptimizerBase *&pRet)
virtual void	SetRegularRate (float f)
	gradient optimizer property
virtual float	GetRegularRate (void)
	get
virtual void	SetSearchLength (float f)
	set
virtual float	GetSearchLength (void)
	get
virtual void	SetRegularMinStepLength (float f)
	set
virtual float	GetRegularMinStepLength (void)
	get
virtual void	SetRegriddingPortion (float f)
	set
virtual float	GetRegriddingPortion (void)
	get
virtual void	Run ()
	for running registration to search optimal
virtual float	AdvanceAlongGradient (zxhTransformBase *pGradientInfCur)
	return no. of steps advanced (may advance fractional steps in future)
virtual void	SetSearchOptimalAlongGradient (int i)
virtual int	GetSearchOptimalAlongGradient ()
virtual void	SetMaxLineSearchStep (int n)
virtual int	GetMaxLineSearchStep ()
virtual bool	AdjustParameters (bool forceconcatenation=false)
virtual bool	ConcatenateTransforms ()
	this is fluid regridding or ( forward composition in localmatrix)
virtual bool	ConcatenateFinalTransforms ()
virtual bool	ConcatenateTransformsNUpdate ()
Protected Member Functions
virtual float	ComputeAndUpdateConjugateDirection (zxhTransformBase *gk_1, zxhTransformBase *gk, zxhTransformBase *dk_1, zxhTransformBase *dk)
	g(k-1),g(k), d(k-1), d(k) from TMI, vol16,no12,pp2879
Protected Attributes
zxhGradientBase *	m_pGradient
float	m_fRegularRate
float	m_fSearchLength
float	m_fRegularMinStepLength
float	m_fRegriddingPortion
int	m_iSearchOptimalAlongGradient
int	m_nMaxLineSearchStep

---

Detailed Description

optimizers search for the optimal value of the metric within the transformation follow gradient direction

---

Member Function Documentation

bool zxhOptimizerGradient::AdjustParameters

(

bool

forceconcatenation = false

)

[virtual]

adjust transform, metric, gradient, differentialstep, ...., Concatenation 1) normal FFDs one2one or ZXHTO test jacobian 2) directional FFDs (set m_pConcatenatedTransformsByRegridding ) return whether regridding

Reimplemented in zxhOptimizerGradientLocalAffines.

zxhOptimizerBase * zxhOptimizerGradient::Clone

(

zxhOptimizerBase *&

pRet

)

[virtual]

Returns:

Reimplemented from zxhOptimizerBase.

Reimplemented in zxhOptimizerGradientField, zxhOptimizerGradientLocalAffines, and zxhOptimizerGradientLocalMatrix.

std::string zxhOptimizerGradient::GetPrintString

(

)

[virtual]

Returns:

Reimplemented from zxhOptimizerBase.

Reimplemented in zxhOptimizerGradientLocalAffines.

virtual int zxhOptimizerGradient::GetSearchOptimalAlongGradient

(

)

[inline, virtual]

0 only advance one step --- defaul 1 advance one step and then forward search until maximal without setting m_bStop 2 search optimal along the gradient and set m_bStop if curr is optimum

void zxhOptimizerGradient::Run

(

)

[virtual]

for running registration to search optimal

1. check to stop if non-increase for some steps

2. compute gradient

3. regular gradient ascent

5. advance along gradient

4. normalise gradient for heart reg

Implements zxhOptimizerBase.

Reimplemented in zxhOptimizerGradientField, and zxhOptimizerGradientLocalAffines.

virtual void zxhOptimizerGradient::SetSearchOptimalAlongGradient

(

int

i

)

[inline, virtual]

0 only advance one step --- defaul 1 advance one step and then forward search until maximal without setting m_bStop 2 search optimal along the gradient and set m_bStop if curr is optimum

---

Member Data Documentation

float zxhOptimizerGradient::m_fRegriddingPortion [protected]

Returns:

float zxhOptimizerGradient::m_fRegularMinStepLength [protected]

Returns:

float zxhOptimizerGradient::m_fRegularRate [protected]

Returns:

float zxhOptimizerGradient::m_fSearchLength [protected]

Returns:

int zxhOptimizerGradient::m_iSearchOptimalAlongGradient [protected]

0 only advance one step --- defaul 1 advance one step and then forward search until maximal without setting m_bStop 2 search optimal along the gradient and set m_bStop if curr is optimum

---

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

• zxhOptimizerGradient.h
• zxhOptimizerGradient.cpp