Calculate masked normalized cross correlation using FFTs. More...

#include <itkMaskedFFTNormalizedCorrelationImageFilter.h>

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

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

Public Types
typedef SmartPointer< const Self >	ConstPointer
typedef FFTImageType::Pointer	FFTImagePointer
typedef Image< std::complex < RealPixelType > , ImageDimension >	FFTImageType
typedef InputImageType::ConstPointer	InputImageConstPointer
typedef InputImageType::Pointer	InputImagePointer
typedef TInputImage	InputImageType
typedef InputImageType::RegionType	InputRegionType
typedef InputImageType::SizeType	InputSizeType
typedef OutputImageType::Pointer	OutputImagePointer
typedef TOutputImage	OutputImageType
typedef OutputImageType::PixelType	OutputPixelType
typedef SmartPointer< Self >	Pointer
typedef RealImageType::Pointer	RealImagePointer
typedef Image< RealPixelType, ImageDimension >	RealImageType
typedef RealImageType::IndexType	RealIndexType
typedef OutputPixelType	RealPixelType
typedef RealImageType::RegionType	RealRegionType
typedef RealImageType::SizeType	RealSizeType
typedef MaskedFFTNormalizedCorrelationImageFilter	Self
typedef ImageToImageFilter < TInputImage, TOutputImage >	Superclass
Public Member Functions
virtual ::itk::LightObject::Pointer	CreateAnother (void) const
virtual const char *	GetNameOfClass () const
	typedef (Concept::IsFloatingPoint< OutputPixelType >) OutputPixelTypeIsFloatingPointCheck

void	SetFixedImage (InputImageType *input)
InputImageType *	GetFixedImage ()

void	SetMovingImage (InputImageType *input)
InputImageType *	GetMovingImage ()

void	SetFixedImageMask (InputImageType *input)
InputImageType *	GetFixedImageMask ()

void	SetMovingImageMask (InputImageType *input)
InputImageType *	GetMovingImageMask ()

virtual void	SetRequiredNumberOfOverlappingVoxels (unsigned long _arg)
virtual unsigned long	GetRequiredNumberOfOverlappingVoxels ()
Static Public Member Functions
static Pointer	New ()
Static Public Attributes
static const unsigned int	ImageDimension = TOutputImage::ImageDimension
Protected Member Functions
template<class LocalInputImageType , class LocalOutputImageType >
LocalOutputImageType::Pointer	CalculateForwardFFT (LocalInputImageType *inputImage, InputSizeType &FFTImageSize)
template<class LocalInputImageType , class LocalOutputImageType >
LocalOutputImageType::Pointer	CalculateInverseFFT (LocalInputImageType *inputImage, RealSizeType &combinedImageSize)
template<class LocalInputImageType >
double	CalculatePrecisionTolerance (LocalInputImageType *inputImage)
template<class LocalInputImageType >
LocalInputImageType::Pointer	ElementPositive (LocalInputImageType *inputImage)
template<class LocalInputImageType , class LocalOutputImageType >
LocalOutputImageType::Pointer	ElementProduct (LocalInputImageType inputImage1, LocalInputImageType inputImage2)
template<class LocalInputImageType >
LocalInputImageType::Pointer	ElementQuotient (LocalInputImageType inputImage1, LocalInputImageType inputImage2)
template<class LocalInputImageType , class LocalOutputImageType >
LocalOutputImageType::Pointer	ElementRound (LocalInputImageType *inputImage)
template<class LocalInputImageType >
LocalInputImageType::Pointer	ElementSubtraction (LocalInputImageType inputImage1, LocalInputImageType inputImage2)
int	FactorizeNumber (int n)
int	FindClosestValidDimension (int n)
void	GenerateData ()
virtual void	GenerateInputRequestedRegion ()
void	GenerateOutputInformation ()
	MaskedFFTNormalizedCorrelationImageFilter ()
template<class LocalInputImageType >
LocalInputImageType::Pointer	PreProcessImage (const LocalInputImageType inputImage, LocalInputImageType inputMask)
template<class LocalInputImageType >
LocalInputImageType::Pointer	PreProcessMask (const LocalInputImageType inputImage, const LocalInputImageType inputMask)
void	PrintSelf (std::ostream &os, Indent indent) const
TInputImage::Pointer	RotateImage (InputImageType *inputImage)
void	VerifyInputInformation ()
virtual	~MaskedFFTNormalizedCorrelationImageFilter ()
Private Member Functions
	MaskedFFTNormalizedCorrelationImageFilter (const Self &)
void	operator= (const Self &)
Private Attributes
unsigned long	m_RequiredNumberOfOverlappingVoxels

Detailed Description

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

Calculate masked normalized cross correlation using FFTs.

This filter calculates the masked normalized cross correlation (NCC) of two images under masks using FFTs instead of spatial correlation. It is much faster than spatial correlation for reasonably large structuring elements. This filter is not equivalent to simply masking the images first and then correlating them; the latter approach yields incorrect results because the zeros in the images still affect the metric in the correlation process. This filter implements the masked NCC correctly so that the masked-out regions are completely ignored. The fundamental difference is described in detail in the references below. If the masks are set to images of all ones, the result of this filter is the same as standard NCC.

Inputs: Two images are required as inputs, fixedImage and movingImage, and two are optional, fixedMask and movingMask. In the context of correlation, inputs are often defined as: "image" and "template". In this filter, the fixedImage plays the role of the image, and the movingImage plays the role of the template. However, this filter is capable of correlating any two images and is not restricted to small movingImages (templates). In the fixedMask and movingMask, non-zero positive values indicate locations of useful information in the corresponding image, whereas zero and negative values indicate locations that should be masked out (ignored). Internally, the masks are converted to have values of only 0 and 1. For each optional mask that is not set, the filter internally creates an image of ones, which is equivalent to not masking the image. Thus, if both masks are not set, the result will be equivalent to unmasked NCC. For example, if only a mask for the fixed image is needed, the movingMask can either not be set or can be set to an image of ones.

Optional parameters: The RequiredNumberOfOverlappingVoxels enables the user to specify how many voxels of the two images must overlap; any location in the correlation map that results from fewer than this number of voxels will be set to zero. Larger values zero-out pixels on a larger border around the correlation image. Thus, larger values remove less stable computations but also limit the capture range. If RequiredNumberOfOverlappingVoxels is set to 0, the default, no zeroing will take place.

Image size: fixedImage and movingImage need not be the same size, but fixedMask must be the same size as fixedImage, and movingMask must be the same size as movingImage. Furthermore, whereas some algorithms require that the "template" be smaller than the "image" because of errors in the regions where the two are not fully overlapping, this filter has no such restriction.

Image spacing: Since the computations are done in the pixel domain, all input images must have the same spacing.

Outputs; The output is an image of RealPixelType that is the masked NCC of the two images and its values range from -1.0 to 1.0. The size of this NCC image is, by definition, size(fixedImage) + size(movingImage) - 1.

Example filter usage:

 typedef itk::MaskedFFTNormalizedCorrelationImageFilter< ShortImageType, DoubleImageType > FilterType;
 FilterType::Pointer filter = FilterType::New();
 filter->SetFixedImage( fixedImage );
 filter->SetMovingImage( movingImage );
 filter->SetFixedImageMask( fixedMask );
 filter->SetMovingImageMask( movingMask );
 filter->SetRequiredNumberOfOverlappingVoxels(20);
 filter->Update();

Warning:: The pixel type of the output image must be of real type (float or double). ConceptChecking is used to enforce the output pixel type. You will get a compilation error if the pixel type of the output image is not float or double.

References: 1) D. Padfield. "Masked object registration in the Fourier domain." Transactions on Image Processing. 2) D. Padfield. "Masked FFT registration". In Proc. Computer Vision and Pattern Recognition, 2010.

Author:: : Dirk Padfield, GE Global Research, padfield@research.ge.com

Definition at line 117 of file itkMaskedFFTNormalizedCorrelationImageFilter.h.