A class in the GeneralisedPrior hierarchy. This implements 'generalised' priors a la the Median Root Prior (which was invented by Sakari Alenius). More...

#include "stir/recon_buildblock/FilterRootPrior.h"

Inheritance diagram for stir::FilterRootPrior< DataT >:

[legend]

Public Member Functions
	FilterRootPrior ()
	Default constructor (no filter)

	FilterRootPrior (shared_ptr< DataProcessor< DataT >> const &, const float penalization_factor)
	Constructs it explicitly.

bool	is_convex () const override
	Indicates if the prior is a smooth convex function. More...

double	compute_value (const DataT &current_estimate) override
	compute the value of the function More...

void	compute_gradient (DataT &prior_gradient, const DataT &current_estimate) override
	compute gradient by applying the filter

Succeeded	set_up (shared_ptr< const DataT > const &target_sptr) override
	Has to be called before using this object.

Public Member Functions inherited from stir::RegisteredParsingObject< FilterRootPrior< DataT >, GeneralisedPrior< DataT >, GeneralisedPrior< DataT > >
std::string	get_registered_name () const override
	Returns Derived::registered_name.

std::string	parameter_info () override
	Returns a string with all parameters and their values, in a form suitable for parsing again.

Public Member Functions inherited from stir::GeneralisedPrior< DataT >
virtual void	compute_Hessian (DataT &prior_Hessian_for_single_densel, const BasicCoordinate< 3, int > &coords, const DataT &current_image_estimate) const
	This computes a single row of the Hessian. More...

virtual void	add_multiplication_with_approximate_Hessian (DataT &output, const DataT &input) const
	This should compute the multiplication of the Hessian with a vector and add it to output. More...

virtual void	accumulate_Hessian_times_input (DataT &output, const DataT &current_estimate, const DataT &input) const
	This should compute the multiplication of the Hessian with a vector and add it to output. More...

float	get_penalisation_factor () const

void	set_penalisation_factor (float new_penalisation_factor)

virtual Succeeded	set_up (shared_ptr< const DataT > const &target_sptr)
	Has to be called before using this object.

Public Member Functions inherited from stir::ParsingObject
	ParsingObject (const ParsingObject &)

ParsingObject &	operator= (const ParsingObject &)

void	ask_parameters ()

bool	parse (std::istream &f)

bool	parse (const char *const filename)

Static Public Attributes
static const char *const	registered_name = "FilterRootPrior"
	Name which will be used when parsing a GeneralisedPrior object.

Protected Member Functions
void	check (DataT const &current_image_estimate) const override
	Check that the prior is ready to be used.

Protected Member Functions inherited from stir::GeneralisedPrior< DataT >
void	set_defaults () override
	sets value for penalisation factor More...

void	initialise_keymap () override
	sets key for penalisation factor More...

virtual void	check (DataT const &current_estimate) const
	Check that the prior is ready to be used.

Protected Member Functions inherited from stir::ParsingObject
virtual bool	post_processing ()
	This will be called at the end of the parsing. More...

virtual void	set_key_values ()
	This will be called before parsing or parameter_info is called. More...

Additional Inherited Members
Static Public Member Functions inherited from stir::RegisteredParsingObject< FilterRootPrior< DataT >, GeneralisedPrior< DataT >, GeneralisedPrior< DataT > >
static GeneralisedPrior< DataT > *	read_from_stream (std::istream *)
	Construct a new object (of type Derived) by parsing the istream. More...

Static Public Member Functions inherited from stir::RegisteredObject< GeneralisedPrior< DataT > >
static GeneralisedPrior< DataT > *	read_registered_object (std::istream *in, const std::string &registered_name)
	Construct a new object (of a type derived from Root, its actual type determined by the registered_name parameter) by parsing the istream. More...

static GeneralisedPrior< DataT > *	ask_type_and_parameters ()
	ask the user for the type, and then calls read_registered_object(0, type) More...

static void	list_registered_names (std::ostream &stream)
	List all possible registered names to the stream. More...

Protected Types inherited from stir::RegisteredObject< GeneralisedPrior< DataT > >
typedef GeneralisedPrior< DataT > (	RootFactory) (std::istream *)
	The type of a root factory is a function, taking an istream* as argument, and returning a Root*.

typedef FactoryRegistry< std::string, RootFactory, interfile_less >	RegistryType
	The type of the registry.

Static Protected Member Functions inherited from stir::RegisteredObject< GeneralisedPrior< DataT > >
static RegistryType &	registry ()
	Static function returning the registry. More...

Protected Attributes inherited from stir::GeneralisedPrior< DataT >
float	penalisation_factor

bool	_already_set_up

Protected Attributes inherited from stir::ParsingObject
KeyParser	parser

Detailed Description

template<typename DataT>
class stir::FilterRootPrior< DataT >

A class in the GeneralisedPrior hierarchy. This implements 'generalised' priors a la the Median Root Prior (which was invented by Sakari Alenius).

This class takes an DataProcessor object (i.e. a filter), and computes the prior 'gradient' as

$G_v = \beta ( {\lambda_v \over F_v} - 1)$

where $\lambda$ is the data where to compute the gradient, and $F$ is the data obtained by filtering $\lambda$ .

However, we need to avoid division by 0, as it might cause a NaN or an 'infinity'. So, we replace the quotient above by
if $|\lambda_v| < M*|F_v|$ then ${\lambda_v \over F_v}$ else $M*\rm{sign}(F_v)*\rm{sign}(lambda_v)$ ,
where $M$ is an arbitrary threshold on the quotient (fixed to 1000 when I wrote this documentation, but check FilterRootPrior.cxx if you want to be sure).

Note that for nearly all filters, this is not a real prior, as this 'gradient' is not the gradient of a function. This can be checked by computing the 'Hessian' (i.e. the partial derivatives of the components of the gradient). For most (interesting) filters, the Hessian will no be symmetric.

The Median Root Prior is obtained by using a MedianImageFilter3D as DataProcessor.

Member Function Documentation

◆ is_convex()

template<typename DataT >

bool stir::FilterRootPrior< DataT >::is_convex ( ) const

overridevirtual

Indicates if the prior is a smooth convex function.

If true, the prior is expected to have 0th, 1st and 2nd order behaviour implemented.

Implements stir::GeneralisedPrior< DataT >.

◆ compute_value()

template<typename DataT >

double stir::FilterRootPrior< DataT >::compute_value ( const DataT & current_estimate )

overridevirtual

compute the value of the function

Warning: Generally there is no function associated to this prior, so we just return 0 and write a warning the first time it's called.

Implements stir::GeneralisedPrior< DataT >.

References stir::warning().

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

/home/sirfuser/devel/STIRdistrib/STIR/src/include/stir/recon_buildblock/FilterRootPrior.h
/home/sirfuser/devel/STIRdistrib/STIR/src/recon_buildblock/FilterRootPrior.cxx

Public Member Functions

Static Public Attributes

Protected Member Functions

Additional Inherited Members

Detailed Description

template<typename DataT> class stir::FilterRootPrior< DataT >

Member Function Documentation

◆ is_convex()

◆ compute_value()

template<typename DataT>
class stir::FilterRootPrior< DataT >