LandmarkBasedTransformInitializer is a helper class intended to The class computes the transform that aligns the fixed and moving images given a set of landmarks. The class is templated over the Transform type. The transform computed gives the best fit transform that maps the fixed and moving images in a least squares sense. The indices are taken to correspond, so point 1 in the first set will get mapped close to point 1 in the second set, etc. An equal number of fixed and moving landmarks need to be specified using SetFixedLandmarks() SetMovingLandmarks(). Any number of landmarks may be specified. Call InitializeTransform() to initialize the transform. More...

#include <itkLandmarkBasedTransformInitializer.h>

Inheritance diagram for itk::LandmarkBasedTransformInitializer< TTransform, TFixedImage, TMovingImage >:

Collaboration diagram for itk::LandmarkBasedTransformInitializer< TTransform, TFixedImage, TMovingImage >:

List of all members.

Public Types
typedef AffineTransform < ParameterValueType, FixedImageType::ImageDimension >	AffineTransformType
typedef SmartPointer< const Self >	ConstPointer
typedef FixedImageType::ConstPointer	FixedImagePointer
typedef TFixedImage	FixedImageType
typedef TransformType::InputPointType	InputPointType
typedef std::vector < LandmarkPointType >	LandmarkPointContainer
typedef Point< double, itkGetStaticConstMacro(ImageDimension) >	LandmarkPointType
typedef LandmarkWeightType::const_iterator	LandmarkWeightConstIterator
typedef std::vector< double >	LandmarkWeightType
typedef MovingImageType::ConstPointer	MovingImagePointer
typedef TMovingImage	MovingImageType
typedef TransformType::OutputVectorType	OutputVectorType
typedef TransformType::ParametersType	ParametersType
typedef ParametersType::ValueType	ParameterValueType
typedef SmartPointer< Self >	Pointer
typedef LandmarkPointContainer::const_iterator	PointsContainerConstIterator
typedef Rigid2DTransform < ParameterValueType >	Rigid2DTransformType
typedef LandmarkBasedTransformInitializer	Self
typedef Object	Superclass
typedef TransformType::Pointer	TransformPointer
typedef TTransform	TransformType
typedef VersorRigid3DTransform < ParameterValueType >	VersorRigid3DTransformType
Public Member Functions
virtual ::itk::LightObject::Pointer	CreateAnother (void) const
virtual const char *	GetNameOfClass () const
virtual void	InitializeTransform ()
void	SetFixedLandmarks (const LandmarkPointContainer &fixedLandmarks)
void	SetLandmarkWeight (LandmarkWeightType &landmarkWeight)
void	SetMovingLandmarks (const LandmarkPointContainer &movingLandmarks)
virtual void	SetTransform (TransformType *_arg)
Static Public Member Functions
static Pointer	New ()
Static Public Attributes
static const unsigned int	ImageDimension = FixedImageType::ImageDimension

static const unsigned int	InputSpaceDimension = TransformType::InputSpaceDimension
static const unsigned int	OutputSpaceDimension = TransformType::OutputSpaceDimension
Protected Types
enum	InputTransformType { VersorRigid3Dtransform = 1, Rigid2Dtransfrom, Else }
Protected Member Functions
	LandmarkBasedTransformInitializer ()
void	PrintSelf (std::ostream &os, Indent indent) const
	~LandmarkBasedTransformInitializer ()
Private Member Functions
template<class TTransform2 >
void	InternalInitializeTransform (TTransform *)
void	InternalInitializeTransform (VersorRigid3DTransformType *)
void	InternalInitializeTransform (Rigid2DTransformType *)
void	InternalInitializeTransform (AffineTransformType *)
	LandmarkBasedTransformInitializer (const Self &)
void	operator= (const Self &)
Private Attributes
FixedImagePointer	m_FixedImage
LandmarkPointContainer	m_FixedLandmarks
LandmarkWeightType	m_LandmarkWeight
MovingImagePointer	m_MovingImage
LandmarkPointContainer	m_MovingLandmarks
TransformPointer	m_Transform

Detailed Description

template<class TTransform, class TFixedImage, class TMovingImage>
class itk::LandmarkBasedTransformInitializer< TTransform, TFixedImage, TMovingImage >

LandmarkBasedTransformInitializer is a helper class intended to The class computes the transform that aligns the fixed and moving images given a set of landmarks. The class is templated over the Transform type. The transform computed gives the best fit transform that maps the fixed and moving images in a least squares sense. The indices are taken to correspond, so point 1 in the first set will get mapped close to point 1 in the second set, etc. An equal number of fixed and moving landmarks need to be specified using SetFixedLandmarks() SetMovingLandmarks(). Any number of landmarks may be specified. Call InitializeTransform() to initialize the transform.

Currently, the following transforms are supported by the class: VersorRigid3DTransform Rigid2DTransform

The class is based in part on Hybrid/vtkLandmarkTransform originally implemented in python by David G. Gobbi.

The solution is based on Berthold K. P. Horn (1987), "Closed-form solution of absolute orientation using unit quaternions," http://people.csail.mit.edu/bkph/papers/Absolute_Orientation.pdf

The Affine Transform initializer is based on an algorithm by H Spaeth, and is described in the Insight Journal Article "Affine Transformation for Landmark Based Registration Initializer in ITK" by Kim E.Y., Johnson H., Williams N. available at http://midasjournal.com/browse/publication/825

Wiki Examples:

All Examples

Rigidly register one image to another using manually specified landmarks

Definition at line 70 of file itkLandmarkBasedTransformInitializer.h.