Computes the inverse of a displacement field. More...

#include <itkInverseDisplacementFieldImageFilter.h>

Inheritance diagram for itk::InverseDisplacementFieldImageFilter< TInputImage, TOutputImage >:

Collaboration diagram for itk::InverseDisplacementFieldImageFilter< TInputImage, TOutputImage >:

Public Types
typedef SmartPointer< const Self >	ConstPointer
typedef OutputImageType::IndexType	IndexType
typedef InputImageType::ConstPointer	InputImageConstPointer
typedef InputImageType::Pointer	InputImagePointer
typedef InputImageType::RegionType	InputImageRegionType
typedef TInputImage	InputImageType
typedef KernelTransformType::Pointer	KernelTransformPointerType
typedef KernelTransform < double, itkGetStaticConstMacro(ImageDimension) >	KernelTransformType
typedef TOutputImage::PointType	OriginPointType
typedef OutputImageType::Pointer	OutputImagePointer
typedef TOutputImage::RegionType	OutputImageRegionType
typedef TOutputImage	OutputImageType
typedef OutputPixelType::ValueType	OutputPixelComponentType
typedef TOutputImage::PixelType	OutputPixelType
typedef SmartPointer< Self >	Pointer
typedef InverseDisplacementFieldImageFilter	Self
typedef OutputImageType::SizeType	SizeType
typedef TOutputImage::SpacingType	SpacingType
typedef ImageToImageFilter < TInputImage, TOutputImage >	Superclass
Public Member Functions
virtual ::itk::LightObject::Pointer	CreateAnother (void) const
virtual void	GenerateInputRequestedRegion ()
virtual void	GenerateOutputInformation ()
virtual KernelTransformType *	GetKernelTransform ()
unsigned long	GetMTime (void) const
virtual const char *	GetNameOfClass () const
virtual const OriginPointType &	GetOutputOrigin ()
virtual const SpacingType &	GetOutputSpacing ()
virtual const SizeType &	GetSize ()
virtual void	SetKernelTransform (KernelTransformType *_arg)
virtual void	SetSize (SizeType _arg)
	typedef (Concept::HasNumericTraits< OutputPixelComponentType >) OutputHasNumericTraitsCheck

virtual void	SetOutputSpacing (SpacingType _arg)
virtual void	SetOutputSpacing (const double *values)

virtual void	SetOutputOrigin (OriginPointType _arg)
virtual void	SetOutputOrigin (const double *values)

virtual void	SetSubsamplingFactor (unsigned int _arg)
virtual unsigned int	GetSubsamplingFactor () const
Static Public Member Functions
static Pointer	New ()
Static Public Attributes
static const unsigned int	ImageDimension = TOutputImage::ImageDimension
Protected Member Functions
void	GenerateData ()
void	PrepareKernelBaseSpline ()

	InverseDisplacementFieldImageFilter ()
	~InverseDisplacementFieldImageFilter ()
void	PrintSelf (std::ostream &os, Indent indent) const
Private Member Functions
	InverseDisplacementFieldImageFilter (const Self &)
void	operator= (const Self &)
Private Attributes
KernelTransformPointerType	m_KernelTransform
OriginPointType	m_OutputOrigin
SpacingType	m_OutputSpacing
SizeType	m_Size
unsigned int	m_SubsamplingFactor

Detailed Description

template<class TInputImage, class TOutputImage>
class itk::InverseDisplacementFieldImageFilter< TInputImage, TOutputImage >

Computes the inverse of a displacement field.

InverseDisplacementFieldImageFilter takes a displacement field as input and computes the displacement field that is its inverse. If the input displacement field was mapping coordinates from a space A into a space B, the output of this filter will map coordinates from the space B into the space A.

Given that both the input and output displacement field are represented as discrete images with pixel type vector, the inverse will be only an estimation and will probably not correspond to a perfect inverse. The precision of the inverse can be improved at the price of increasing the computation time and memory consumption in this filter.

The method used for computing the inverse displacement field is to subsample the input field using a regular grid and create Kerned-Base Spline in which the reference landmarks are the coordinates of the deformed point and the target landmarks are the negative of the displacement vectors. The kernel-base spline is then used for regularly sampling the output space and recover vector values for every single pixel.

The subsampling factor used for the regular grid of the input field will determine the number of landmarks in the KernelBased spline and therefore it will have a dramatic effect on both the precision of output displacement field and the computational time required for the filter to complete the estimation. A large subsampling factor will result in few landmarks in the KernelBased spline, therefore on fast computation and low precision. A small subsampling factor will result in a large number of landmarks in the KernelBased spline, therefore a large memory consumption, long computation time and high precision for the inverse estimation.

This filter expects both the input and output images to be of pixel type Vector.

Definition at line 65 of file itkInverseDisplacementFieldImageFilter.h.