ITK
6.0.0
Insight Toolkit
|
#include <itkMRFImageFilter.h>
Implementation of a labeler object that uses Markov Random Fields to classify pixels in an image data set.
This object classifies pixels based on a Markov Random Field (MRF) model. This implementation uses the maximum a posteriori (MAP) estimates for modeling the MRF. The object traverses the data set and uses the model generated by the Mahalanobis distance classifier to get the distance between each pixel in the data set to a set of known classes, updates the distances by evaluating the influence of its neighboring pixels (based on a MRF model) and finally, classifies each pixel to the class which has the minimum distance to that pixel (taking the neighborhood influence under consideration). DoNeighborhoodOperation is the function that can be modified to achieve different flavors of MRF filters in derived classes.
The classified initial labeled image is needed. It is important that the number of expected classes be set before calling the classifier. In our case we have used the ImageClassifier using a Gaussian model to generate the initial labels. This classifier requires the user to ensure that an appropriate membership function is provided. See the documentation of the image classifier class for more information.
The influence of a neighborhood on a given pixel's classification (the MRF term) is computed by calculating a weighted sum of number of class labels in a three dimensional neighborhood. The basic idea of this neighborhood influence is that if a large number of neighbors of a pixel are of one class, then the current pixel is likely to be of the same class.
The dimensions of the neighborhood are the same as the input image dimensions, and values of the weighting parameters are either specified by the user or through the beta matrix parameter. The default weighting table is generated during object construction. The following table shows an example of a 3 x 3 x 3 neighborhood and the weighting values used. A 3 x 3 x 3 kernel is used where each value is a nonnegative parameter, which encourages neighbors to be of the same class. In this example, the influence of the pixels in the same slice is assigned a weight 1.7. The influence of the pixels in the same location in the previous and next slice is assigned a weight 1.5. A weight of 1.3 is assigned to the influence of the north, south, east, west, and diagonal pixels in the previous and next slices.
\[\begin{tabular}{ccc} \begin{tabular}{|c|c|c|} 1.3 & 1.3 & 1.3 \\ 1.3 & 1.5 & 1.3 \\ 1.3 & 1.3 & 1.3 \\ \end{tabular} & \begin{tabular}{|c|c|c|} 1.7 & 1.7 & 1.7 \\ 1.7 & 0 & 1.7 \\ 1.7 & 1.7 & 1.7 \\ \end{tabular} & \begin{tabular}{|c|c|c|} 1.3 & 1.3 & 1.3 \\ 1.5 & 1.5 & 1.3 \\ 1.3 & 1.3 & 1.3 \\ \end{tabular} \\ \end{tabular}\]
The user needs to set the neighborhood size using the SetNeighborhoodRadius function. The details on the semantics of a neighborhood can be found in the documentation associated with the itkNeighborhood and related objects. NOTE: The size of the neighborhood must match the size of the neighborhood weighting parameters set by the user.
For minimization of the MRF labeling function the MinimizeFunctional virtual method is called. For our current implementation we use the iterated conditional modes (ICM) algorithm described by Besag in the paper "On the Statistical Analysis of Dirty Pictures" in J. Royal Stat. Soc. B, Vol. 48, 1986.
In each iteration, the algorithm visits each pixel in turn and determines whether to update its classification by computing the influence of the classification of the pixel's neighbors and of the intensity data. On each iteration after the first, we reexamine the classification of a pixel only if the classification of some of its neighbors has changed in the previous iteration. The pixels' classification is updated using a synchronous scheme (iteration by iteration) until the error reaches less than the threshold or the number of iteration exceed the maximum set number of iterations. Note: The current implementation supports betaMatrix default weight for two and three dimensional images only. The default for higher dimension is set to unity. This should be overridden by custom weights after filter initialization.
Definition at line 149 of file itkMRFImageFilter.h.
Static Public Member Functions | |
static Pointer | New () |
Static Public Member Functions inherited from itk::ImageToImageFilter< TInputImage, TClassifiedImage > | |
static double | GetGlobalDefaultCoordinateTolerance () |
static double | GetGlobalDefaultDirectionTolerance () |
static void | SetGlobalDefaultCoordinateTolerance (double) |
static void | SetGlobalDefaultDirectionTolerance (double) |
Static Public Member Functions inherited from itk::Object | |
static bool | GetGlobalWarningDisplay () |
static void | GlobalWarningDisplayOff () |
static void | GlobalWarningDisplayOn () |
static Pointer | New () |
static void | SetGlobalWarningDisplay (bool val) |
Static Public Member Functions inherited from itk::LightObject | |
static void | BreakOnError () |
static Pointer | New () |
Static Public Attributes | |
static constexpr unsigned int | ClassifiedImageDimension = TClassifiedImage::ImageDimension |
static constexpr unsigned int | InputImageDimension = TInputImage::ImageDimension |
Static Public Attributes inherited from itk::ImageToImageFilter< TInputImage, TClassifiedImage > | |
static constexpr unsigned int | InputImageDimension |
static constexpr unsigned int | OutputImageDimension |
Static Public Attributes inherited from itk::ImageSource< TClassifiedImage > | |
static constexpr unsigned int | OutputImageDimension |
Protected Types | |
using | LabelStatusImageIterator = ImageRegionIterator< LabelStatusImageType > |
using | LabelStatusImageNeighborhoodIterator = NeighborhoodIterator< LabelStatusImageType > |
using | LabelStatusImagePointer = typename LabelStatusImageType::Pointer |
using | LabelStatusImageType = Image< int, Self::InputImageDimension > |
using | LabelStatusIndexType = typename LabelStatusImageType::IndexType |
using | LabelStatusRegionType = typename LabelStatusImageType::RegionType |
Protected Types inherited from itk::ImageToImageFilter< TInputImage, TClassifiedImage > | |
using | InputToOutputRegionCopierType = ImageToImageFilterDetail::ImageRegionCopier< Self::OutputImageDimension, Self::InputImageDimension > |
using | OutputToInputRegionCopierType = ImageToImageFilterDetail::ImageRegionCopier< Self::InputImageDimension, Self::OutputImageDimension > |
Private Types | |
using | InputImageSizeType = typename TInputImage::SizeType |
using | LabelStatusImageFaceListIterator = typename LabelStatusImageFaceListType::iterator |
using | LabelStatusImageFaceListType = typename LabelStatusImageFacesCalculator::FaceListType |
using | LabelStatusImageFacesCalculator = NeighborhoodAlgorithm::ImageBoundaryFacesCalculator< LabelStatusImageType > |
using | LabelStatusImageNeighborhoodRadiusType = typename LabelStatusImageNeighborhoodIterator::RadiusType |
Private Member Functions | |
void | ApplyICMLabeller () |
virtual void | SetDefaultMRFNeighborhoodWeight () |
Private Attributes | |
ClassifierType::Pointer | m_ClassifierPtr {} |
double * | m_ClassProbability { nullptr } |
std::vector< double > | m_DummyVector {} |
int | m_ErrorCounter { 0 } |
double | m_ErrorTolerance { 0.2 } |
InputImageNeighborhoodRadiusType | m_InputImageNeighborhoodRadius {} |
unsigned int | m_KernelSize {} |
LabelledImageNeighborhoodRadiusType | m_LabelledImageNeighborhoodRadius {} |
LabelStatusImagePointer | m_LabelStatusImage {} |
LabelStatusImageNeighborhoodRadiusType | m_LabelStatusImageNeighborhoodRadius {} |
std::vector< double > | m_MahalanobisDistance {} |
unsigned int | m_MaximumNumberOfIterations { 50 } |
std::vector< double > | m_MRFNeighborhoodWeight {} |
int | m_NeighborhoodSize { 27 } |
std::vector< double > | m_NeighborInfluence {} |
unsigned int | m_NumberOfClasses { 0 } |
unsigned int | m_NumberOfIterations { 0 } |
double | m_SmoothingFactor { 1 } |
MRFStopEnum | m_StopCondition { MRFStopEnum::MaximumNumberOfIterations } |
int | m_TotalNumberOfPixelsInInputImage { 1 } |
int | m_TotalNumberOfValidPixelsInOutputImage { 1 } |
Additional Inherited Members | |
Static Protected Member Functions inherited from itk::ImageSource< TClassifiedImage > | |
static const ImageRegionSplitterBase * | GetGlobalDefaultSplitter () |
static ITK_THREAD_RETURN_FUNCTION_CALL_CONVENTION | ThreaderCallback (void *arg) |
Static Protected Member Functions inherited from itk::ProcessObject | |
template<typename TSourceObject > | |
static void | MakeRequiredOutputs (TSourceObject &sourceObject, const DataObjectPointerArraySizeType numberOfRequiredOutputs) |
static constexpr float | progressFixedToFloat (uint32_t fixed) |
static uint32_t | progressFloatToFixed (float f) |
Protected Attributes inherited from itk::ImageSource< TClassifiedImage > | |
bool | m_DynamicMultiThreading |
Protected Attributes inherited from itk::ProcessObject | |
TimeStamp | m_OutputInformationMTime {} |
bool | m_Updating {} |
Protected Attributes inherited from itk::LightObject | |
std::atomic< int > | m_ReferenceCount {} |
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::ClassifierType = ImageClassifierBase<TInputImage, TClassifiedImage> |
Type definitions for classifier to be used for the MRF labeling.
Definition at line 218 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::ConstPointer = SmartPointer<const Self> |
Definition at line 159 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::IndexValueType = typename LabelledImageIndexType::IndexValueType |
Definition at line 206 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::InputImageConstPointer = typename TInputImage::ConstPointer |
Definition at line 171 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::InputImageFaceListIterator = typename InputImageFaceListType::iterator |
Definition at line 235 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::InputImageFaceListType = typename InputImageFacesCalculator::FaceListType |
Definition at line 233 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::InputImageFacesCalculator = NeighborhoodAlgorithm::ImageBoundaryFacesCalculator<TInputImage> |
Definition at line 231 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::InputImageNeighborhoodIterator = ConstNeighborhoodIterator<TInputImage> |
Input image neighborhood iterator and kernel size type alias
Definition at line 227 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::InputImageNeighborhoodRadiusType = typename InputImageNeighborhoodIterator::RadiusType |
Definition at line 229 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::InputImagePixelType = typename TInputImage::PixelType |
Type definition for the input image pixel type.
Definition at line 174 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::InputImagePointer = typename TInputImage::Pointer |
Definition at line 170 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::InputImageRegionConstIterator = ImageRegionConstIterator<TInputImage> |
Definition at line 181 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::InputImageRegionIterator = ImageRegionIterator<TInputImage> |
Type definition for the input image region iterator
Definition at line 180 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::InputImageRegionType = typename TInputImage::RegionType |
Type definition for the input image region type.
Definition at line 177 of file itkMRFImageFilter.h.
|
private |
Definition at line 393 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::InputImageType = TInputImage |
Type definition for the input image.
Definition at line 169 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::LabelledImageFaceListIterator = typename LabelledImageFaceListType::iterator |
Definition at line 246 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::LabelledImageFaceListType = typename LabelledImageFacesCalculator::FaceListType |
Definition at line 244 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::LabelledImageFacesCalculator = NeighborhoodAlgorithm::ImageBoundaryFacesCalculator<TClassifiedImage> |
Definition at line 242 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::LabelledImageIndexType = typename TClassifiedImage::IndexType |
Type definition for the classified image index type.
Definition at line 205 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::LabelledImageNeighborhoodIterator = NeighborhoodIterator<TClassifiedImage> |
Labeled image neighborhood iterator type alias
Definition at line 238 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::LabelledImageNeighborhoodRadiusType = typename LabelledImageNeighborhoodIterator::RadiusType |
Definition at line 240 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::LabelledImageOffsetType = typename TClassifiedImage::OffsetType |
Type definition for the classified image offset type.
Definition at line 209 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::LabelledImagePixelType = typename TClassifiedImage::PixelType |
Type definitions for the classified image pixel type. It has to be the same type as the training image.
Definition at line 198 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::LabelledImagePointer = typename TClassifiedImage::Pointer |
Type definitions for the labelled image. It is derived from the training image.
Definition at line 194 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::LabelledImageRegionIterator = ImageRegionIterator<TClassifiedImage> |
Type definition for the input image region iterator
Definition at line 212 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::LabelledImageRegionType = typename TClassifiedImage::RegionType |
Type definitions for the classified image pixel type. It has to be the same type as the training image.
Definition at line 202 of file itkMRFImageFilter.h.
|
private |
Definition at line 401 of file itkMRFImageFilter.h.
|
private |
Definition at line 399 of file itkMRFImageFilter.h.
|
private |
Definition at line 397 of file itkMRFImageFilter.h.
|
protected |
Definition at line 369 of file itkMRFImageFilter.h.
|
protected |
Labelled status image neighborhood iterator type alias
Definition at line 372 of file itkMRFImageFilter.h.
|
private |
Definition at line 395 of file itkMRFImageFilter.h.
|
protected |
Definition at line 368 of file itkMRFImageFilter.h.
|
protected |
Definition at line 365 of file itkMRFImageFilter.h.
|
protected |
Definition at line 366 of file itkMRFImageFilter.h.
|
protected |
Definition at line 367 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::MRFStopEnum = MRFImageFilterEnums::MRFStop |
Definition at line 316 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::NeighborhoodRadiusType = typename TInputImage::SizeType |
Radius type alias support
Definition at line 224 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::Pointer = SmartPointer<Self> |
Definition at line 158 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::Self = MRFImageFilter |
Standard class type aliases.
Definition at line 156 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::SizeType = typename TInputImage::SizeType |
Size and value type alias support
Definition at line 221 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::Superclass = ImageToImageFilter<TInputImage, TClassifiedImage> |
Definition at line 157 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::TrainingImagePixelType = typename TClassifiedImage::PixelType |
Type definitions for the training image pixel type.
Definition at line 190 of file itkMRFImageFilter.h.
using itk::MRFImageFilter< TInputImage, TClassifiedImage >::TrainingImagePointer = typename TClassifiedImage::Pointer |
Type definitions for the training image.
Definition at line 187 of file itkMRFImageFilter.h.
|
protected |
|
overrideprotecteddefault |
|
protected |
Allocate memory for labelled images.
|
private |
Apply the ICM algorithm to label the images.
|
protectedvirtual |
Apply MRF Classifier. In this example the images are labelled using Iterated Conditional Mode algorithm by J. Besag, "On statistical analysis of dirty pictures," J. Royal Stat. Soc. B, vol. 48, pp. 259-302, 1986.
|
protectedvirtual |
Perform the MRF operation with each neighborhood.
|
overrideprotectedvirtual |
Give the process object a chance to indicate that it will produce more output than it was requested to produce. For example, many imaging filters must compute the entire output at once or can only produce output in complete slices. Such filters cannot handle smaller requested regions. These filters must provide an implementation of this method, setting the output requested region to the size they will produce. By default, a process object does not modify the size of the output requested region.
Reimplemented from itk::ProcessObject.
|
overrideprotectedvirtual |
This method causes the filter to generate its output.
Reimplemented from itk::ProcessObject.
Reimplemented in itk::RGBGibbsPriorFilter< TInputImage, TClassifiedImage >.
|
overrideprotectedvirtual |
What is the input requested region that is required to produce the output requested region? By default, the largest possible region is always required but this is overridden in many subclasses. For instance, for an image processing filter where an output pixel is a simple function of an input pixel, the input requested region will be set to the output requested region. For an image processing filter where an output pixel is a function of the pixels in a neighborhood of an input pixel, then the input requested region will need to be larger than the output requested region (to avoid introducing artificial boundary conditions). This function should never request an input region that is outside the the input largest possible region (i.e. implementations of this method should crop the input requested region at the boundaries of the input largest possible region).
Reimplemented from itk::ProcessObject.
|
overrideprotectedvirtual |
Generate the information describing the output data. The default implementation of this method will copy information from the input to the output. A filter may override this method if its output will have different information than its input. For instance, a filter that shrinks an image will need to provide an implementation for this method that changes the spacing of the pixels. Such filters should call their superclass' implementation of this method prior to changing the information values they need (i.e. GenerateOutputInformation() should call Superclass::GenerateOutputInformation() prior to changing the information.
Reimplemented from itk::ProcessObject.
|
virtual |
Set/Get the error tolerance level which is used as a threshold to quit the iterations
|
virtual |
Set/Get the number of iteration of the Iterated Conditional Mode (ICM) algorithm. A default value is set at 50 iterations.
Reimplemented in itk::RGBGibbsPriorFilter< TInputImage, TClassifiedImage >.
|
inlinevirtual |
Definition at line 311 of file itkMRFImageFilter.h.
|
overridevirtual |
Reimplemented from itk::ProcessObject.
Reimplemented in itk::RGBGibbsPriorFilter< TInputImage, TClassifiedImage >.
|
inline |
Get the neighborhood radius
Definition at line 291 of file itkMRFImageFilter.h.
|
virtual |
Set/Get the number of classes.
Reimplemented in itk::RGBGibbsPriorFilter< TInputImage, TClassifiedImage >.
|
virtual |
|
virtual |
Set/Get the degree of smoothing desired
|
virtual |
Get condition that stops the MRF filter (Number of Iterations / Error tolerance )
|
protectedvirtual |
Minimize the functional to be used.
Reimplemented in itk::RGBGibbsPriorFilter< TInputImage, TClassifiedImage >.
|
static |
Method for creation through the object factory.
|
overrideprotectedvirtual |
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::ProcessObject.
Reimplemented in itk::RGBGibbsPriorFilter< TInputImage, TClassifiedImage >.
void itk::MRFImageFilter< TInputImage, TClassifiedImage >::SetClassifier | ( | typename ClassifierType::Pointer | ptrToClassifier | ) |
Set the pointer to the classifier being used.
|
privatevirtual |
Set/Get the weighting parameters (Beta Matrix). A default 3 x 3 x 3 matrix is provided. However, the user is allowed to override it with their choice of weights for a 3 x 3 x 3 matrix.
|
virtual |
Set/Get the error tolerance level which is used as a threshold to quit the iterations
|
virtual |
Set/Get the number of iteration of the Iterated Conditional Mode (ICM) algorithm. A default value is set at 50 iterations.
Reimplemented in itk::RGBGibbsPriorFilter< TInputImage, TClassifiedImage >.
|
virtual |
Set the weighting parameters (used in MRF algorithms). This is a function allowing the users to set the weight matrix by providing a a 1D array of weights. The default implementation supports a 3 x 3 x 3 kernel. The labeler needs to be extended for a different kernel size.
void itk::MRFImageFilter< TInputImage, TClassifiedImage >::SetNeighborhoodRadius | ( | const NeighborhoodRadiusType & | ) |
Set the neighborhood radius.
void itk::MRFImageFilter< TInputImage, TClassifiedImage >::SetNeighborhoodRadius | ( | const SizeValueType * | radiusArray | ) |
void itk::MRFImageFilter< TInputImage, TClassifiedImage >::SetNeighborhoodRadius | ( | const | SizeValueType | ) |
Sets the radius for the neighborhood.
Calculates size from the radius, and allocates storage.
|
virtual |
Set/Get the number of classes.
Reimplemented in itk::RGBGibbsPriorFilter< TInputImage, TClassifiedImage >.
|
virtual |
Set/Get the degree of smoothing desired
|
staticconstexpr |
Labelled Image dimension
Definition at line 215 of file itkMRFImageFilter.h.
|
staticconstexpr |
Image dimension
Definition at line 184 of file itkMRFImageFilter.h.
|
private |
Pointer to the classifier to be used for the MRF labelling.
Definition at line 429 of file itkMRFImageFilter.h.
|
private |
Definition at line 417 of file itkMRFImageFilter.h.
|
private |
Definition at line 426 of file itkMRFImageFilter.h.
|
private |
Definition at line 411 of file itkMRFImageFilter.h.
|
private |
Definition at line 415 of file itkMRFImageFilter.h.
|
private |
Definition at line 403 of file itkMRFImageFilter.h.
|
private |
Definition at line 409 of file itkMRFImageFilter.h.
|
private |
Definition at line 404 of file itkMRFImageFilter.h.
|
private |
Definition at line 421 of file itkMRFImageFilter.h.
|
private |
Definition at line 405 of file itkMRFImageFilter.h.
|
private |
Definition at line 425 of file itkMRFImageFilter.h.
|
private |
Definition at line 408 of file itkMRFImageFilter.h.
|
private |
Definition at line 423 of file itkMRFImageFilter.h.
|
private |
Definition at line 412 of file itkMRFImageFilter.h.
|
private |
Definition at line 424 of file itkMRFImageFilter.h.
|
private |
Definition at line 407 of file itkMRFImageFilter.h.
|
private |
Definition at line 418 of file itkMRFImageFilter.h.
|
private |
Definition at line 416 of file itkMRFImageFilter.h.
|
private |
Definition at line 419 of file itkMRFImageFilter.h.
|
private |
Definition at line 414 of file itkMRFImageFilter.h.
|
private |
Definition at line 413 of file itkMRFImageFilter.h.