ITK  4.0.0
Insight Segmentation and Registration Toolkit
Public Types | Public Member Functions | Static Public Member Functions | Static Public Attributes | Protected Member Functions | Private Member Functions | Private Attributes
itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet > Class Template Reference

Implementation of the Jensen Havrda Charvat Tsallis Point Set metric. More...

#include <itkJensenHavrdaCharvatTsallisPointSetToPointSetMetricv4.h>

Inheritance diagram for itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >:
Collaboration diagram for itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >:

List of all members.

Public Types

typedef SmartPointer< const SelfConstPointer
typedef Superclass::CoordRepType CoordRepType
typedef
DensityFunctionType::Pointer 
DensityFunctionPointer
typedef
ManifoldParzenWindowsPointSetFunction
< PointSetType, RealType
DensityFunctionType
typedef Superclass::DerivativeType DerivativeType
typedef
DensityFunctionType::GaussianType 
GaussianType
typedef
Superclass::LocalDerivativeType 
LocalDerivativeType
typedef Superclass::MeasureType MeasureType
typedef
Superclass::NeighborsIdentifierType 
NeighborsIdentifierType
typedef SmartPointer< SelfPointer
typedef Superclass::PointIdentifier PointIdentifier
typedef
PointSetType::PointsContainer 
PointsContainer
typedef
PointsContainer::ConstIterator 
PointsContainerConstIterator
typedef TPointSet PointSetType
typedef Superclass::PointType PointType
typedef MeasureType RealType
typedef
JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4 
Self
typedef
PointSetToPointSetMetricv4
< TPointSet, TPointSet > 
Superclass

Public Member Functions

virtual ::itk::LightObject::Pointer CreateAnother (void) const
virtual RealType GetAlpha () const
virtual unsigned int GetCovarianceKNeighborhood () const
virtual void GetDerivative (DerivativeType &) const
virtual unsigned int GetEvaluationKNeighborhood () const
virtual RealType GetKernelSigma () const
virtual void GetLocalNeighborhoodValueAndDerivative (const PointType &, MeasureType &, LocalDerivativeType &) const
virtual const char * GetNameOfClass () const
virtual RealType GetPointSetSigma () const
virtual bool GetUseAnisotropicCovariances () const
virtual bool GetUseRegularizationTerm () const
virtual MeasureType GetValue () const
virtual void GetValueAndDerivative (MeasureType &, DerivativeType &) const
virtual void Initialize (void) throw ( ExceptionObject )
virtual void SetAlpha (RealType _arg)
virtual void SetCovarianceKNeighborhood (unsigned int _arg)
virtual void SetEvaluationKNeighborhood (unsigned int _arg)
virtual void SetKernelSigma (RealType _arg)
virtual void SetPointSetSigma (RealType _arg)
virtual void SetUseAnisotropicCovariances (bool _arg)
virtual void SetUseRegularizationTerm (bool _arg)
virtual void UseAnisotropicCovariancesOff ()
virtual void UseAnisotropicCovariancesOn ()
virtual void UseRegularizationTermOff ()
virtual void UseRegularizationTermOn ()
virtual MeasureType GetLocalNeighborhoodValue (const PointType &) const

Static Public Member Functions

static Pointer New ()

Static Public Attributes

static const unsigned int PointDimension = TPointSet::PointDimension

Protected Member Functions

 JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4 ()
void PrintSelf (std::ostream &os, Indent indent) const
 ~JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4 ()

Private Member Functions

 JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4 (const Self &)
void operator= (const Self &)

Private Attributes

RealType m_Alpha
unsigned int m_CovarianceKNeighborhood
unsigned int m_EvaluationKNeighborhood
DensityFunctionPointer m_FixedDensityFunction
RealType m_KernelSigma
DensityFunctionPointer m_MovingDensityFunction
RealType m_PointSetSigma
bool m_UseAnisotropicCovariances
bool m_UseRegularizationTerm

Detailed Description

template<class TPointSet>
class itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >

Implementation of the Jensen Havrda Charvat Tsallis Point Set metric.

Given a specified transform and direction, this class calculates the value and derivative between a "fixed" and "moving" point set pair using the Havrda-Charvat-Tsallis entropy family, a generalization of the well-known Shannon entropy, and the Jensen divergence. Another way to look at the family of information-theoretic measures is that the points are used to construct the corresponding probably density functions.

In addition, we allow the user to invoke a manifold parzen windowing of the data. Instead of an isotropic Gaussian being associated with each point, we can actually calculate the covariance matrix for each point such that it reflects the locate point set structure.

To speed up the metric calculation, we use ITK's K-d tree to query the metric value only for a given neighborhood. Considering that probably only a small subset of points is needed to get a good approximation of the metric value for a single point, this is probably warranted. So what we do is transform each point (with the specified transform) and construct the k-d tree from the transformed points.

Contributed by Nicholas J. Tustison, James C. Gee in the Insight Journal paper: http://hdl.handle.net/1926/1524

REFERENCE

N.J. Tustison, S. P. Awate, G. Song, T. S. Cook, and J. C. Gee. "Point set registration using Havrda-Charvat-Tsallis entropy measures" IEEE Transactions on Medical Imaging, 30(2):451-60, 2011.

Definition at line 62 of file itkJensenHavrdaCharvatTsallisPointSetToPointSetMetricv4.h.


Member Typedef Documentation

template<class TPointSet >
typedef SmartPointer<const Self> itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::ConstPointer

Other typedefs

Definition at line 99 of file itkJensenHavrdaCharvatTsallisPointSetToPointSetMetricv4.h.

Type of the derivative.

Reimplemented from itk::PointSetToPointSetMetricv4< TPointSet, TPointSet >.

Definition at line 86 of file itkJensenHavrdaCharvatTsallisPointSetToPointSetMetricv4.h.

Types transferred from the base class

Reimplemented from itk::PointSetToPointSetMetricv4< TPointSet, TPointSet >.

Definition at line 85 of file itkJensenHavrdaCharvatTsallisPointSetToPointSetMetricv4.h.

template<class TPointSet >
typedef SmartPointer<Self> itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::Pointer
template<class TPointSet >
typedef PointSetType::PointsContainer itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::PointsContainer
template<class TPointSet >
typedef PointsContainer::ConstIterator itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::PointsContainerConstIterator
template<class TPointSet >
typedef TPointSet itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::PointSetType
template<class TPointSet >
typedef MeasureType itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::RealType

Standard class typedefs.

Reimplemented from itk::PointSetToPointSetMetricv4< TPointSet, TPointSet >.

Definition at line 67 of file itkJensenHavrdaCharvatTsallisPointSetToPointSetMetricv4.h.

template<class TPointSet >
typedef PointSetToPointSetMetricv4<TPointSet, TPointSet> itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::Superclass

Constructor & Destructor Documentation


Member Function Documentation

template<class TPointSet >
virtual::itk::LightObject::Pointer itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::CreateAnother ( void  ) const [virtual]

Create an object from an instance, potentially deferring to a factory. This method allows you to create an instance of an object that is exactly the same type as the referring object. This is useful in cases where an object has been cast back to a base class.

Reimplemented from itk::Object.

template<class TPointSet >
virtual RealType itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::GetAlpha ( ) const [virtual]

Get the alpha parameter used to tune the point-set metric.

template<class TPointSet >
virtual unsigned int itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::GetCovarianceKNeighborhood ( ) const [virtual]

Get the size of the covariance neighborhood used to construct the anisotropic covariances. Only relevant if m_UseAnisotropicCovariances = true. Default = 5.

template<class TPointSet >
virtual void itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::GetDerivative ( DerivativeType ) const [virtual]

This method returns the derivative based on the current transformation(s).

Reimplemented from itk::PointSetToPointSetMetricv4< TPointSet, TPointSet >.

template<class TPointSet >
virtual unsigned int itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::GetEvaluationKNeighborhood ( ) const [virtual]

Get the neighborhood size used to evaluate the measurement at each point. Default = 50.

template<class TPointSet >
virtual RealType itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::GetKernelSigma ( ) const [virtual]

Get the noise kernel sigma for the anistropic covariances.

template<class TPointSet >
virtual MeasureType itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::GetLocalNeighborhoodValue ( const PointType ) const [inline, virtual]

Pure virtual function from the parent class not needed in this class

Definition at line 223 of file itkJensenHavrdaCharvatTsallisPointSetToPointSetMetricv4.h.

References itkExceptionMacro.

template<class TPointSet >
virtual void itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::GetLocalNeighborhoodValueAndDerivative ( const PointType ,
MeasureType ,
LocalDerivativeType  
) const [inline, virtual]

Pure virtual function from the parent class not needed in this class

Definition at line 231 of file itkJensenHavrdaCharvatTsallisPointSetToPointSetMetricv4.h.

References itkExceptionMacro.

template<class TPointSet >
virtual const char* itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::GetNameOfClass ( ) const [virtual]

Run-time type information (and related methods)

Reimplemented from itk::PointSetToPointSetMetricv4< TPointSet, TPointSet >.

template<class TPointSet >
virtual RealType itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::GetPointSetSigma ( ) const [virtual]

Get the point set sigma function

template<class TPointSet >
virtual bool itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::GetUseAnisotropicCovariances ( ) const [virtual]

Get whether or not anisotropic covariances are determined for each Gaussian. Default = false.

template<class TPointSet >
virtual bool itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::GetUseRegularizationTerm ( ) const [virtual]

Get boolean usage of regularization term.

template<class TPointSet >
virtual MeasureType itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::GetValue ( ) const [virtual]

This method returns the value of the metric based on the current transformation(s).

Reimplemented from itk::PointSetToPointSetMetricv4< TPointSet, TPointSet >.

template<class TPointSet >
virtual void itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::GetValueAndDerivative ( MeasureType ,
DerivativeType  
) const [virtual]

This method returns the derivative and value based on the current transformation(s).

Reimplemented from itk::PointSetToPointSetMetricv4< TPointSet, TPointSet >.

template<class TPointSet >
virtual void itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::Initialize ( void  ) throw ( ExceptionObject ) [virtual]

Initialize the Metric by making sure that all the components are present and plugged together correctly

Reimplemented from itk::PointSetToPointSetMetricv4< TPointSet, TPointSet >.

template<class TPointSet >
static Pointer itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::New ( ) [static]

Method for creation through the object factory.

Reimplemented from itk::Object.

template<class TPointSet >
void itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::operator= ( const Self ) [private]

Mutex lock to protect modification to the reference count

Reimplemented from itk::PointSetToPointSetMetricv4< TPointSet, TPointSet >.

template<class TPointSet >
void itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::PrintSelf ( std::ostream &  os,
Indent  indent 
) const [protected, virtual]

Methods invoked by Print() to print information about the object including superclasses. Typically not called by the user (use Print() instead) but used in the hierarchical print process to combine the output of several classes.

Reimplemented from itk::PointSetToPointSetMetricv4< TPointSet, TPointSet >.

template<class TPointSet >
virtual void itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::SetAlpha ( RealType  _arg) [virtual]

Set the alpha parameter used to tune the point-set metric from a maximum-likelihood measure (alpha = 1) to the more robust L2 solution (alpha = 2). Typically, "robustness" is associated with performance in the presence of uniform noise but in most applications the noise will be highly correlated with the point sets, therefore an alpha value close to 1, in general, provides better performance. Only values between 1 and 2 are convex.

template<class TPointSet >
virtual void itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::SetCovarianceKNeighborhood ( unsigned int  _arg) [virtual]

Set the size of the covariance neighborhood used to construct the anisotropic covariances. Only relevant if m_UseAnisotropicCovariances = true. Default = 5.

template<class TPointSet >
virtual void itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::SetEvaluationKNeighborhood ( unsigned int  _arg) [virtual]

Set the neighborhood size used to evaluate the measurement at each point. Default = 50.

template<class TPointSet >
virtual void itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::SetKernelSigma ( RealType  _arg) [virtual]

Set the size of the noise kernel used to construct each covariance image. To avoid the case where the local point set structure would result in a degenerate covariance matrix, a small amount of noise is added along the diagonal represented by this variable. Only relevant if m_UseAnisotropicCovariances = true. Default = 10.0.

template<class TPointSet >
virtual void itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::SetPointSetSigma ( RealType  _arg) [virtual]

Each point is associated with a Gaussian characterized by m_PointSetSigma which provides a sense of scale for determining the similarity between two point sets. Default = 1.0.

template<class TPointSet >
virtual void itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::SetUseAnisotropicCovariances ( bool  _arg) [virtual]

Set whether or not anisotropic covariances are determined for each Gaussian. Default = false.

template<class TPointSet >
virtual void itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::SetUseRegularizationTerm ( bool  _arg) [virtual]

The Jensen divergence is comprised of two terms---one measures the mutual similarity while the second provides a regularizing effect penalizing tendencies towards a single point. For most registration applications, a separate transform would be used which would provide its own regularization. Default = false.

template<class TPointSet >
virtual void itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::UseAnisotropicCovariancesOff ( ) [virtual]
template<class TPointSet >
virtual void itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::UseAnisotropicCovariancesOn ( ) [virtual]

Get/set whether or not anisotropic covariances are determined for each Gaussian. Default = false.

template<class TPointSet >
virtual void itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::UseRegularizationTermOff ( ) [virtual]
template<class TPointSet >
virtual void itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::UseRegularizationTermOn ( ) [virtual]

Get/set boolean usage of regularization term.


Member Data Documentation

template<class TPointSet >
RealType itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::m_Alpha [private]
template<class TPointSet >
unsigned int itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::m_CovarianceKNeighborhood [private]
template<class TPointSet >
unsigned int itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::m_EvaluationKNeighborhood [private]
template<class TPointSet >
RealType itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::m_KernelSigma [private]
template<class TPointSet >
RealType itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::m_PointSetSigma [private]
template<class TPointSet >
bool itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::m_UseAnisotropicCovariances [private]
template<class TPointSet >
bool itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::m_UseRegularizationTerm [private]
template<class TPointSet >
const unsigned int itk::JensenHavrdaCharvatTsallisPointSetToPointSetMetricv4< TPointSet >::PointDimension = TPointSet::PointDimension [static]

typedefs for the data types used in the point set metric calculations. It is assumed that the constants of the fixed point set, such as the point dimension, are the same for the "common space" in which the metric calculation occurs.

Reimplemented from itk::PointSetToPointSetMetricv4< TPointSet, TPointSet >.

Definition at line 82 of file itkJensenHavrdaCharvatTsallisPointSetToPointSetMetricv4.h.


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