Gruen pattern matching algorithm. More...

#include <Gruen.h>

Inheritance diagram for Isis::Gruen:

Collaboration diagram for Isis::Gruen:

Public Types
enum	RegisterStatus { SuccessPixel , SuccessSubPixel , PatternChipNotEnoughValidData , FitChipNoData , FitChipToleranceNotMet , SurfaceModelNotEnoughValidData , SurfaceModelSolutionInvalid , SurfaceModelDistanceInvalid , PatternZScoreNotMet , AdaptiveAlgorithmFailed }
	Enumeration of the Register() method's return status. More...

enum	GradientFilterType { None , Sobel }
	Enumeration of the different types of gradient filters that can be applied to the pattern and search chips before matching them. More...

Public Member Functions
	Gruen ()
	Default constructor.

	Gruen (Pvl &pvl)
	Construct a Gruen search algorithm.

virtual	~Gruen ()
	Destructor.

BigInt	CallCount () const
	Returns the current call count.

void	WriteSubsearchChips (const QString &pattern="SubChip")
	Set up for writing subsearch for a a given registration call.

AffineTolerance	getAffineTolerance () const
	Return set of tolerances for affine convergence.

double	getSpiceConstraint () const
	Returns the SPICE tolerance constraint as read from config file.

double	getAffineConstraint () const
	Returns the Affine tolerance constraint as read from config file.

void	setAffineRadio (const AffineRadio &affrad)
	Sets initial chip transformation.

void	setAffineRadio ()
	Set affine parameters to defaults.

const AffineRadio &	getDefaultAffineRadio () const
	Returns default settings for Affine/Radiometric parameters.

const AffineRadio &	getAffineRadio () const
	Return current state of Affine/Radio state

MatchPoint	getLastMatch () const
	Returns the register state of the last successful Gruen match.

Chip *	PatternChip ()
	Return pointer to pattern chip.

Chip *	SearchChip ()
	Return pointer to search chip.

Chip *	FitChip ()
	Return pointer to fit chip.

Chip *	RegistrationPatternChip ()
	Return pointer to pattern chip used in registration.

Chip *	RegistrationSearchChip ()
	Return pointer to search chip used in registration.

Chip *	ReducedPatternChip ()
	Return pointer to reduced pattern chip.

Chip *	ReducedSearchChip ()
	Return pointer to reduced search chip.

Chip *	ReducedFitChip ()
	Return pointer to reduced fit chip.

void	SetSubPixelAccuracy (bool on)
	If the sub-pixel accuracy is enabled, the Register() method will attempt to match the pattern chip to the search chip at sub-pixel accuracy, otherwise it will be registered at whole pixel accuracy.

void	SetPatternValidPercent (const double percent)
	Set the amount of data in the pattern chip that must be valid.

void	SetSubsearchValidPercent (const double percent)
	Set the amount of data in the search chip's subchip that must be valid.

void	SetTolerance (double tolerance)
	Set the tolerance for an acceptable goodness of fit.

void	SetChipInterpolator (const QString &interpolator)
	Sets the Chip class interpolator type to be used to load pattern and search chips.

void	SetSurfaceModelWindowSize (int size)
	Set the surface model window size.

void	SetSurfaceModelDistanceTolerance (double distance)
	Set a distance the surface model solution is allowed to move away from the best whole pixel fit in the fit chip.

void	SetReductionFactor (int reductionFactor)
	Set the reduction factor used to speed up the pattern matching algorithm.

void	SetPatternZScoreMinimum (double minimum)
	Set the minimum pattern zscore.

void	SetGradientFilterType (const QString &gradientFilterType)
	Set the gradient filter type to be applied to the search and pattern chips.

QString	GradientFilterString () const

bool	SubPixelAccuracy ()
	Return whether this object will attempt to register to whole or sub-pixel accuracy.

int	ReductionFactor ()
	Return the reduction factor.

double	PatternValidPercent () const
	Return pattern chip valid percent. The default value is.

double	SubsearchValidPercent () const
	Return subsearch chip valid percent.

double	Tolerance () const
	Return match algorithm tolerance.

double	WindowSize () const
	Return window size.

double	DistanceTolerance () const
	Return distance tolerance.

void	Distance (double &sampDistance, double &lineDistance)
	Return the distance point moved.

AutoReg::RegisterStatus	Register ()
	Walk the pattern chip through the search chip to find the best registration.

bool	Success () const
	Return whether the match algorithm succeeded or not.

double	GoodnessOfFit () const
	Return the goodness of fit of the match algorithm.

bool	IsIdeal (double fit)
	Returns true if the fit parameter is arbitrarily close to the ideal fit value.

double	ChipSample () const
	Return the search chip sample that best matched.

double	ChipLine () const
	Return the search chip line that best matched.

double	CubeSample () const
	Return the search chip cube sample that best matched.

double	CubeLine () const
	Return the search chip cube line that best matched.

double	MinimumZScore () const
	Return minimumPatternZScore.

void	ZScores (double &score1, double &score2) const
	Return the ZScores of the pattern chip.

Pvl	RegistrationStatistics ()
	This returns the cumulative registration statistics.

virtual double	MostLenientTolerance ()
	Minimum tolerance specific to algorithm.

PvlGroup	RegTemplate ()
	This function returns the keywords that this object was created from.

PvlGroup	UpdatedTemplate ()
	Returns a PvlGroup containing the PvlKeywords of the parameters this object was most recently run with.

Protected Member Functions
virtual QString	AlgorithmName () const
	Returns the default name of the algorithm as Gruen.

int	algorithm (Chip &pattern, Chip &subsearch, const Radiometric &radio, BigInt &ptsUsed, double &resid, GMatrix &atai, AffineRadio &affrad)
	Real workhorse of the computational Gruen algorithm.

Analysis	errorAnalysis (const BigInt &npts, const double &resid, const GMatrix &atai)
	Compute the error analysis of convergent Gruen matrix.

virtual double	MatchAlgorithm (Chip &pattern, Chip &subsearch)
	Minimization of data set using Gruen algorithm.

virtual bool	CompareFits (double fit1, double fit2)
	This virtual method must return if the 1st fit is equal to or better than the second fit.

virtual double	IdealFit () const
	Returns the ideal fit for a perfect Gruen result.

virtual AutoReg::RegisterStatus	Registration (Chip &sChip, Chip &pChip, Chip &fChip, int startSamp, int startLine, int endSamp, int endLine, int bestSamp, int bestLine)
	Applies the adaptive Gruen algorithm to pattern and search chips.

virtual Pvl	AlgorithmStatistics (Pvl &pvl)
	Create Gruen error and processing statistics Pvl output.

void	SetChipSample (double sample)
	Sets the search chip subpixel sample that matches the pattern tack sample.

void	SetChipLine (double line)
	Sets the search chip subpixel line that matches the pattern tack line.

void	SetGoodnessOfFit (double fit)
	Sets the goodness of fit for adaptive algorithms.

void	Parse (Pvl &pvl)
	Initialize parameters in the AutoReg class using a PVL specification.

bool	SetSubpixelPosition (Chip &window)
	Set the search chip sample and line to subpixel values if possible.

Chip	Reduce (Chip &chip, int reductionFactor)
	This method reduces the given chip by the given reduction factor.

Protected Attributes
PvlObject	p_template
	AutoRegistration object that created this projection.

Detailed Description

Gruen pattern matching algorithm.

This class provides adaptive image (chip) registration using the AutoReg factory architecture. This algorithm uses an Affine transform to iteratively adjust the search chip at each iteration. Each iteration solves for new adjustments to the Affine transform until the 6 affine parameters fall below the tolerances as specified in AffineThreshHold1 and AffineThreshHold2.

This class minimizes the 6 specifiable Affine transform components of a 3x3 matrix. The three Affine components for X (sample) and Y (line) are scale, shear and translation. Gruen provides control over the maximum values these three components should attain in order for the registration to converge to a successful match. These limits are specified by AffineScaleTolerance, AffineShearTolerance and AffineTranslationTolerance. AffineShearTolerance is optional and if not specified, it defaults to the value of AffineScaleTolerance. These tolerances specify the maximum amount of translation pixels can be shifted between one Gruen iteration and another. For example, AffineTranslationTolerance = 0.2 means that a subsearch chip cannot move in sample or line direction more than 0.2 pixels in order to satisfy convergence. AffineScaleTolerance constrains the sample and line scale elements of the Affine transformation. And AffineShearTolerance constrains the sample and line shear elements of the Affine transformation. The scale and shear parameters are scaled by size of the Chip. The sample scale and shear Affine components maximum limit is computed as AffineScaleTolerance/((#Samples-1)/2) and AffineShearTolerance/((#Samples-1)/2). Likewise, the line scale and shear maximums are computed using #Lines in the preceding equation.

See also: MinimumGruen AdaptiveGruen AutoReg

Author: 2009-09-09 Kris Becker

Member Enumeration Documentation

◆ GradientFilterType

enum Isis::AutoReg::GradientFilterType

inherited

Enumeration of the different types of gradient filters that can be applied to the pattern and search chips before matching them.

Enumerator
None	default, no gradient filter
Sobel	Sobel gradient filter.

◆ RegisterStatus

enum Isis::AutoReg::RegisterStatus

inherited

Enumeration of the Register() method's return status.

All of the enumerations describe a failure to register except "Success". These status values can be used to provide the user with more specific feedback on why registration did not succeed.

Enumerator
SuccessPixel	Success registering to whole pixel.
SuccessSubPixel	Success registering to sub-pixel accuracy.
PatternChipNotEnoughValidData	Not enough valid data in pattern chip.
FitChipNoData	Fit chip did not have any valid data.
FitChipToleranceNotMet	Goodness of fit tolerance not satisfied.
SurfaceModelNotEnoughValidData	Not enough points to fit a surface model for sub-pixel accuracy.
SurfaceModelSolutionInvalid	Could not model surface for sub-pixel accuracy.
SurfaceModelDistanceInvalid	Surface model moves registration more than one pixel.
PatternZScoreNotMet	Pattern data max or min does not pass the z-score test.
AdaptiveAlgorithmFailed	Error occured in Adaptive algorithm.

Constructor & Destructor Documentation

◆ Gruen() [1/2]

Isis::Gruen::Gruen ( )

Default constructor.

Default constructor sets up default Gruen parameters.

Author: Kris Becker - 5/22/2011

◆ Gruen() [2/2]

Isis::Gruen::Gruen ( Pvl & pvl )

Construct a Gruen search algorithm.

This will construct a minimum difference search algorith. It is recommended that you use a AutoRegFactory class as opposed to this constructor. Direct construction is used commonly in stereo matching.

Parameters

pvl	A Pvl object that contains a valid automatic registration definition

◆ ~Gruen()

virtual Isis::Gruen::~Gruen ( )

inlinevirtual

Destructor.

Member Function Documentation

◆ algorithm()

int Isis::Gruen::algorithm	(	Chip &	pattern,
		Chip &	subsearch,
		const Radiometric &	radio,
		BigInt &	ptsUsed,
		double &	resid,
		GMatrix &	atai,
		AffineRadio &	affrad )

protected

Real workhorse of the computational Gruen algorithm.

This method is called for all registration requests and actually performs the registration of two chips.

The pattern chip is deemed constant. The subsearch chip is generally an extraction from the search chip that has had an affine transform applied to fill it.

At each iteration of the Gruen algorithm, the affine transform is incrementally updated based upon the results from this method. There are six affine parameters and two radiometric (shift and gain) parameters that are solved/computed here.

The algorithm itself is a first derivative computation of the subsearch chip with small radiometric adjustments applied to better tone match the two chips. This is intended to minimize the affine variability.

Parameters

pattern	Fixed pattern chip which subsearch is trying to match
subsearch	Affined extraction of the search chip

Returns: Returns 0 if successful, otherwise returns the error number associated with the problem encountered (see ErrorTypes)

Referenced by MatchAlgorithm(), and Registration().

◆ AlgorithmName()

virtual QString Isis::Gruen::AlgorithmName ( ) const

inlineprotectedvirtual

Returns the default name of the algorithm as Gruen.

Implements Isis::AutoReg.

Reimplemented in Isis::AdaptiveGruen.

Referenced by AlgorithmStatistics().

◆ AlgorithmStatistics()

Pvl Isis::Gruen::AlgorithmStatistics ( Pvl & pvl )

protectedvirtual

Create Gruen error and processing statistics Pvl output.

This method generates two groups specific to the Gruen algorithm: The GruenFailures group which logs all the errors enountered during processing and the GruenStatistics group which logs selected statistics gathered during a registration run.

These groups are added to the AutoReg log output Pvl container for reporting to user/log files.

Parameters

pvl	Input AutoReg Pvl container to add results to

Returns: Pvl Output Pvl container with Gruen information

Reimplemented from Isis::AutoReg.

References AlgorithmName(), Isis::CollectorMap< K, T, ComparePolicy, RemovalPolicy, CopyPolicy >::getNth(), Isis::CollectorMap< K, T, ComparePolicy, RemovalPolicy, CopyPolicy >::size(), and Isis::toString().

◆ CallCount()

BigInt Isis::Gruen::CallCount ( ) const

inline

Returns the current call count.

◆ ChipLine()

double Isis::AutoReg::ChipLine ( ) const

inlineinherited

Return the search chip line that best matched.

◆ ChipSample()

double Isis::AutoReg::ChipSample ( ) const

inlineinherited

Return the search chip sample that best matched.

◆ CompareFits()

bool Isis::Gruen::CompareFits	(	double	fit1,
		double	fit2 )

protectedvirtual

This virtual method must return if the 1st fit is equal to or better than the second fit.

Parameters

fit1	1st goodness of fit
fit2	2nd goodness of fit

Reimplemented from Isis::AutoReg.

◆ CubeLine()

double Isis::AutoReg::CubeLine ( ) const

inlineinherited

Return the search chip cube line that best matched.

◆ CubeSample()

double Isis::AutoReg::CubeSample ( ) const

inlineinherited

Return the search chip cube sample that best matched.

◆ Distance()

void Isis::AutoReg::Distance	(	double &	sampDistance,
		double &	lineDistance )

inlineinherited

Return the distance point moved.

Parameters

sampDistance	Sample movement
lineDistance	Line movement

◆ DistanceTolerance()

double Isis::AutoReg::DistanceTolerance ( ) const

inlineinherited

Return distance tolerance.

Referenced by Isis::AutoReg::UpdatedTemplate().

◆ errorAnalysis()

Analysis Isis::Gruen::errorAnalysis	(	const BigInt &	npts,
		const double &	resid,
		const GMatrix &	atai )

protected

Compute the error analysis of convergent Gruen matrix.

Author: kbecker (5/15/2011)

Parameters

npts
resid
atai

Returns: Analysis

References Isis::Analysis::m_npts.

Referenced by MatchAlgorithm(), and Registration().

◆ FitChip()

Chip * Isis::AutoReg::FitChip ( )

inlineinherited

Return pointer to fit chip.

Referenced by Isis::ControlMeasureEditWidget::saveChips(), and Isis::ControlPointEdit::saveChips().

◆ getAffineConstraint()

double Isis::Gruen::getAffineConstraint ( ) const

inline

Returns the Affine tolerance constraint as read from config file.

◆ getAffineRadio()

const AffineRadio & Isis::Gruen::getAffineRadio ( ) const

inline

Return current state of Affine/Radio state

◆ getAffineTolerance()

AffineTolerance Isis::Gruen::getAffineTolerance ( ) const

Return set of tolerances for affine convergence.

Author: kbecker (5/15/2011)

Returns: Tolerance Affine tolerances from PVL setup

Referenced by Registration().

◆ getDefaultAffineRadio()

const AffineRadio & Isis::Gruen::getDefaultAffineRadio ( ) const

inline

Returns default settings for Affine/Radiometric parameters.

Referenced by setAffineRadio().

◆ getLastMatch()

MatchPoint Isis::Gruen::getLastMatch ( ) const

inline

Returns the register state of the last successful Gruen match.

This method returns the full match condition of the last call to Gruen Register function that was successful.

BEWARE: This is only valid if Register returns successfully! This is due to AutoReg returning conditions that occur prior to the actual Gruen algorithm being called.

Author: Kris Becker - 6/4/2011

Returns: MatchPoint

◆ getSpiceConstraint()

double Isis::Gruen::getSpiceConstraint ( ) const

inline

Returns the SPICE tolerance constraint as read from config file.

◆ GoodnessOfFit()

double Isis::AutoReg::GoodnessOfFit ( ) const

inlineinherited

Return the goodness of fit of the match algorithm.

◆ GradientFilterString()

QString Isis::AutoReg::GradientFilterString ( ) const

inherited

References _FILEINFO_, Isis::AutoReg::None, Isis::IException::Programmer, and Isis::AutoReg::Sobel.

Referenced by Isis::AutoReg::UpdatedTemplate().

◆ IdealFit()

virtual double Isis::Gruen::IdealFit ( ) const

inlineprotectedvirtual

Returns the ideal fit for a perfect Gruen result.

Implements Isis::AutoReg.

◆ IsIdeal()

bool Isis::AutoReg::IsIdeal ( double fit )

inlineinherited

Returns true if the fit parameter is arbitrarily close to the ideal fit value.

Parameters

fit	Fit value to be compared to the ideal fit

Returns: bool Indicates whether the fit is ideal

References Isis::AutoReg::IdealFit().

Referenced by Isis::AutoReg::Registration().

◆ MatchAlgorithm()

double Isis::Gruen::MatchAlgorithm	(	Chip &	pattern,
		Chip &	subsearch )

protectedvirtual

Minimization of data set using Gruen algorithm.

This is a very minimal application of the Gruen algorithm that provides the ability to use it in a non-adaptive capacity. This method processes two chips of the same size, pattern and subsearch. The subsearch has typically been extracted in the same manner as the MinimumDifference or MaximumCorrelation routines are utilized.

It simply applies the algorithm to the current state of the two chips, computes the error analysis on it and returns the eigen vector solution as an indication of chip registration integrity.

Note that in this mode, most all the parameters found in the definition file that apply to the adaptive mode are ignored when utilizing the algorithm in this fashion.

Parameters

pattern	[in] A Chip object usually containing an nxm area of a cube. Must be the same diminsions as subsearch.
subsearch	[in] A Chip object usually containing an nxm area of a cube. Must be the same dimensions as pattern. This is normally a subarea of a larger portion of the image.

Returns: The square root of the eigen values of DN differences OR Isis::NULL if the Gruen algorithm fails.

Implements Isis::AutoReg.

References algorithm(), errorAnalysis(), and Isis::Null.

◆ MinimumZScore()

double Isis::AutoReg::MinimumZScore ( ) const

inlineinherited

Return minimumPatternZScore.

Referenced by Isis::AutoReg::UpdatedTemplate().

◆ MostLenientTolerance()

virtual double Isis::AutoReg::MostLenientTolerance ( )

inlinevirtualinherited

Minimum tolerance specific to algorithm.

Reimplemented in Isis::MinimumDifference.

◆ Parse()

void Isis::AutoReg::Parse ( Pvl & pvl )

protectedinherited

Initialize parameters in the AutoReg class using a PVL specification.

An example of the PVL required for this is:

Object = AutoRegistration
  Group = Algorithm
    Name      = MaximumCorrelation
    Tolerance = 0.7
  EndGroup
 
  Group = PatternChip
    Samples = 21
    Lines   = 21
  EndGroup
 
  Group = SearchChip
    Samples = 51
    Lines = 51
  EndGroup
EndObject

There are many other options that can be set via the pvl and are described in other documentation (see below).

See also: patternMatch.doc under the coreg application

Parameters

pvl	The pvl object containing the specification

Exceptions

iException::User "Improper format for AutoReg PVL."

Referenced by Isis::AutoReg::AutoReg().

◆ PatternChip()

Chip * Isis::AutoReg::PatternChip ( )

inlineinherited

Return pointer to pattern chip.

Referenced by Isis::AutoReg::Parse(), and Isis::AutoReg::UpdatedTemplate().

◆ PatternValidPercent()

double Isis::AutoReg::PatternValidPercent ( ) const

inlineinherited

Return pattern chip valid percent. The default value is.

Referenced by Isis::MaximumCorrelation::MatchAlgorithm(), and Isis::AutoReg::UpdatedTemplate().

◆ Reduce()

Chip Isis::AutoReg::Reduce	(	Chip &	chip,
		int	reductionFactor )

protectedinherited

This method reduces the given chip by the given reduction factor.

Used to speed up the match algorithm.

Parameters

chip	Chip to be reduced
reductionFactor	Factor by which to reduce chip.

Returns: Chip Reduced chip object

References Isis::Chip::GetValue(), Isis::Chip::Lines(), Isis::Null, Isis::Statistics::Reset(), Isis::Chip::Samples(), and Isis::Chip::SetValue().

Referenced by Isis::AutoReg::Register().

◆ ReducedFitChip()

Chip * Isis::AutoReg::ReducedFitChip ( )

inlineinherited

Return pointer to reduced fit chip.

◆ ReducedPatternChip()

Chip * Isis::AutoReg::ReducedPatternChip ( )

inlineinherited

Return pointer to reduced pattern chip.

◆ ReducedSearchChip()

Chip * Isis::AutoReg::ReducedSearchChip ( )

inlineinherited

Return pointer to reduced search chip.

◆ ReductionFactor()

int Isis::AutoReg::ReductionFactor ( )

inlineinherited

Return the reduction factor.

Referenced by Isis::AutoReg::UpdatedTemplate().

◆ Register()

AutoReg::RegisterStatus Isis::AutoReg::Register ( )

inherited

Walk the pattern chip through the search chip to find the best registration.

Returns: AutoReg::RegisterStatus Returns the status of the registration.

Exceptions

iException::User	- "Search chips samples must be at least N pixels wider than the pattern chip samples for successful surface modeling"
iException::User	- "Search chips lines must be at least N pixels taller than the pattern chip lines for successful surface modeling"
iException::User	- "Reduction factor is too large"

References _FILEINFO_, Isis::AutoReg::FitChipNoData, Isis::Chip::Lines(), Isis::AutoReg::None, Isis::Null, Isis::AutoReg::PatternChipNotEnoughValidData, Isis::AutoReg::PatternZScoreNotMet, Isis::AutoReg::Reduce(), Isis::AutoReg::Registration(), Isis::Chip::Samples(), Isis::Chip::SetChipPosition(), Isis::Chip::SetSize(), Isis::Chip::SetValue(), Isis::AutoReg::Success(), Isis::AutoReg::SuccessSubPixel, and Isis::IException::User.

◆ Registration()

AutoReg::RegisterStatus Isis::Gruen::Registration	(	Chip &	sChip,
		Chip &	pChip,
		Chip &	fChip,
		int	startSamp,
		int	startLine,
		int	endSamp,
		int	endLine,
		int	bestSamp,
		int	bestLine )

protectedvirtual

Applies the adaptive Gruen algorithm to pattern and search chips.

This method computes the adaptive Gruen algorithm for a pattern chip and search chip. The search chip is assumed to be of a larger size than the pattern chip as dictated by the contents of the registration definition file.

This algorithm can be used with or without "fast geoming" the search chip. It works quite well where the two images are assumed to be nearly spatially registered. Its real intent is to compute parallax angles between two images taken at different viewing geometry. This provides an efficient process for deriving a digital elevation model (DEM) from two datasets.

The Gruen algorithm is applied to the chips until the algorithm converges (current iteration yields a detla affine within tolerance limits), an error is encountered, or the maximum number of iterations is exceeded.

Note that bestSamp and bestLine may not be the original center of the search chip. It is subject to chip reduction matching as specified by the user. All distance tolerances are compute from this postion. The process of chip reduction processing is handled by AutoReg prior to calling this routine.

See also: getThreshHold() for Affine limit information

Parameters

sChip	Full search chip as rendered from the search image
pChip	Full pattern chip as rendered from the pattern/match image
fChip	Maintains the solution vector at each chip location
startSamp	Starting sample of the search image range
startLine	Starting line of the search image range
endSamp	Ending sample of the search image range
endLine	Ending line of the search image range
bestSamp	Best registering sample of search chip
bestLine	best registering line of search chip

Returns: AutoReg::RegisterStatus Returns AutoReg::Success if the chip is successfully registered, otherwise returns AutoReg::AdaptiveAlgorithmFailed.

Reimplemented from Isis::AutoReg.

References algorithm(), errorAnalysis(), getAffineTolerance(), Isis::AutoReg::SetChipLine(), Isis::AutoReg::SetChipSample(), Isis::AutoReg::SetGoodnessOfFit(), and Isis::MatchPoint::setStatus().

◆ RegistrationPatternChip()

Chip * Isis::AutoReg::RegistrationPatternChip ( )

inlineinherited

Return pointer to pattern chip used in registration.

References Isis::AutoReg::None.

Referenced by Isis::ControlMeasureEditWidget::saveChips(), and Isis::ControlPointEdit::saveChips().

◆ RegistrationSearchChip()

Chip * Isis::AutoReg::RegistrationSearchChip ( )

inlineinherited

Return pointer to search chip used in registration.

References Isis::AutoReg::None.

Referenced by Isis::ControlMeasureEditWidget::saveChips(), and Isis::ControlPointEdit::saveChips().

◆ RegistrationStatistics()

Pvl Isis::AutoReg::RegistrationStatistics ( )

inherited

This returns the cumulative registration statistics.

That is, the Register() method accumulates statistics regard the errors each time is called. Invoking this method returns a PVL summary of those statisitics

Author: janderson (3/26/2009)

Returns: Pvl

References Isis::AutoReg::AlgorithmStatistics(), and Isis::toString().

◆ RegTemplate()

PvlGroup Isis::AutoReg::RegTemplate ( )

inherited

This function returns the keywords that this object was created from.

Returns: PvlGroup The keywords this object used in initialization

References Isis::PvlObject::findGroup(), Isis::PvlObject::hasGroup(), Isis::AutoReg::p_template, Isis::AutoReg::SetPatternValidPercent(), Isis::AutoReg::SetSubsearchValidPercent(), and Isis::PvlObject::Traverse.

◆ SearchChip()

Chip * Isis::AutoReg::SearchChip ( )

inlineinherited

Return pointer to search chip.

Referenced by Isis::AutoReg::Parse(), and Isis::AutoReg::UpdatedTemplate().

◆ setAffineRadio() [1/2]

void Isis::Gruen::setAffineRadio ( )

Set affine parameters to defaults.

This method differs from the one above in that it uses the defaults as defined at construction. The basic difference is that this call sets the affine portion to the identity and the radiometric parameters to the defaults as provided in the user input auto-regististration parameters. It may have default shift and gain values to use.

Author: Kris Becker - 6/4/2011

References getDefaultAffineRadio().

◆ setAffineRadio() [2/2]

void Isis::Gruen::setAffineRadio ( const AffineRadio & affrad )

Sets initial chip transformation.

This method can be used with AutoReg registration to set initial affine transform parameters. This initial condition will be applied to the whole search chip extraction for the first subsearch chip. The caller must define the contents of the affine and radiometric parameters. See the AffineRadio construct for details.

Author: kbecker (5/14/2011)

Parameters

affrad Initial Affine/Radio parameters to apply on registration entry

◆ SetChipInterpolator()

void Isis::AutoReg::SetChipInterpolator ( const QString & interpolator )

inherited

Sets the Chip class interpolator type to be used to load pattern and search chips.

Acceptable values for the interpolator parameter include:

NearestNeighborType
BiLinearType
CubicConvolutionType

If this method is not called, the chip interpolator type defaults to CubicConvolutionType in the Chip class.

Parameters

interpolator Name of interpolator type to be used. This is taken from the Pvl's ChipInterpolator keyword value.

Exceptions

iException::User - "Invalid Interpolator type."

Author: Jeannie Walldren

References _FILEINFO_, Isis::Interpolator::BiLinearType, Isis::Interpolator::CubicConvolutionType, Isis::Interpolator::NearestNeighborType, Isis::Chip::SetReadInterpolator(), and Isis::IException::User.

Referenced by Isis::AutoReg::Parse().

◆ SetChipLine()

void Isis::AutoReg::SetChipLine ( double line )

inlineprotectedinherited

Sets the search chip subpixel line that matches the pattern tack line.

Parameters

line	Value to set for search chip subpixel line

Referenced by Registration().

◆ SetChipSample()

void Isis::AutoReg::SetChipSample ( double sample )

inlineprotectedinherited

Sets the search chip subpixel sample that matches the pattern tack sample.

Parameters

sample Value to set for search chip subpixel sample

Referenced by Registration().

◆ SetGoodnessOfFit()

void Isis::AutoReg::SetGoodnessOfFit ( double fit )

inlineprotectedinherited

Sets the goodness of fit for adaptive algorithms.

Parameters

fit	Fit value to set

Referenced by Registration().

◆ SetGradientFilterType()

void Isis::AutoReg::SetGradientFilterType ( const QString & gradientFilterType )

inherited

Set the gradient filter type to be applied to the search and pattern chips.

Parameters

gradientFilterType the gradient filter type to use

Exceptions

iException::User - "Invalid Gradient type."

References _FILEINFO_, Isis::AutoReg::None, Isis::AutoReg::Sobel, and Isis::IException::User.

Referenced by Isis::AutoReg::Parse().

◆ SetPatternValidPercent()

void Isis::AutoReg::SetPatternValidPercent ( const double percent )

inherited

Set the amount of data in the pattern chip that must be valid.

For example, a 21x21 pattern chip has 441 pixels. If percent is 75 then at least 330 pixels pairs must be valid in order for a comparision between the pattern and search sub-region to occur. That is, both the pattern pixel and search pixel must be valid to be counted. Pixels are considered valid based on the min/max range specified on each of the Chips (see Chip::SetValidRange method).

If the pattern chip reduction option is used this percentage will apply to all reduced patterns. Additionally, the pattern sampling effects the pixel count. For example if pattern sampling is 50% then only 220 pixels in the 21x21 pattern are considered so 165 must be valid.

If this method is not called, the PatternChip ValidPercent defaults to 50 in the AutoReg object constructor.

See also: SetValidRange()

Parameters

percent Percentage of valid data between 0 and 100, default is 50% if never invoked

Exceptions

iException::User - "Invalid value for PatternChip ValidPercent."

References _FILEINFO_, and Isis::IException::User.

Referenced by Isis::AutoReg::AutoReg(), Isis::AutoReg::Parse(), and Isis::AutoReg::RegTemplate().

◆ SetPatternZScoreMinimum()

void Isis::AutoReg::SetPatternZScoreMinimum ( double minimum )

inherited

Set the minimum pattern zscore.

This option is used to ignore pattern chips which are bland (low standard deviation). If the minimum or maximum pixel value in the pattern chip does not meet the minimum zscore value (see a statisitcs book for definition of zscore) then invalid registration will occur.

If this method is not called, the z-score minimum defaults to 1.0 in the AutoReg object constructor.

Parameters

minimum The minimum zscore value for the pattern chip. Default is 1.0

Exceptions

iException::User - "Invalid value for PatternChip MinimumZScore."

References _FILEINFO_, and Isis::IException::User.

Referenced by Isis::AutoReg::AutoReg(), and Isis::AutoReg::Parse().

◆ SetReductionFactor()

void Isis::AutoReg::SetReductionFactor ( int factor )

inherited

Set the reduction factor used to speed up the pattern matching algorithm.

If this method is not called, the reduction factor defaults to 1 in the AutoReg object constructor.

Parameters

factor Reduction factor. Must be greater than or equal to 1.

Exceptions

iException::User - "Invalid value for Algorithm ReductionFactor."

References _FILEINFO_, and Isis::IException::User.

Referenced by Isis::AutoReg::AutoReg(), and Isis::AutoReg::Parse().

◆ SetSubPixelAccuracy()

void Isis::AutoReg::SetSubPixelAccuracy ( bool on )

inherited

If the sub-pixel accuracy is enabled, the Register() method will attempt to match the pattern chip to the search chip at sub-pixel accuracy, otherwise it will be registered at whole pixel accuracy.

If this method is not called, the sub pixel accuracy defaults to on = true in the AutoReg object constructor.

Parameters

on	Set the state of registration accuracy. The default is sub-pixel accuracy is on

Referenced by Isis::AutoReg::AutoReg(), and Isis::AutoReg::Parse().

◆ SetSubpixelPosition()

bool Isis::AutoReg::SetSubpixelPosition ( Chip & window )

protectedinherited

Set the search chip sample and line to subpixel values if possible.

This method uses a centroiding method to gravitate the whole pixel best fit to a subpixel extremum in the continuous image space. The weights of the centers of gravity in the centroiding algorithm are modeled by goodness of fit values within a discrete search window.

Parameters

window The search window extracted from the fit chip

Returns: bool Returns true if the subpixel solution is valid

References Isis::Centroid::setDNRange().

Referenced by Isis::AutoReg::Registration().

◆ SetSubsearchValidPercent()

void Isis::AutoReg::SetSubsearchValidPercent ( const double percent )

inherited

Set the amount of data in the search chip's subchip that must be valid.

If this method is not called, the SearchChip SubchipValidPercent defaults to 50 in the AutoReg object constructor.

Parameters

percent Percentage of valid data between 0 and 100, default is 50% if never invoked

See also: SetPatternValidPercent()

Exceptions

iException::User - "Invalid value for SearchChip SubchipValidPercent."

References _FILEINFO_, and Isis::IException::User.

Referenced by Isis::AutoReg::AutoReg(), Isis::AutoReg::Parse(), and Isis::AutoReg::RegTemplate().

◆ SetSurfaceModelDistanceTolerance()

void Isis::AutoReg::SetSurfaceModelDistanceTolerance ( double distance )

inherited

Set a distance the surface model solution is allowed to move away from the best whole pixel fit in the fit chip.

If this method is not called, the distance tolerance defaults to 1.5 in the AutoReg object constructor.

Parameters

distance The distance allowed to move in pixels. Must be greater than 0.

Exceptions

iException::User - "Invalid value for SurfaceModel DistanceTolerance."

References _FILEINFO_, and Isis::IException::User.

Referenced by Isis::AutoReg::AutoReg(), and Isis::AutoReg::Parse().

◆ SetSurfaceModelWindowSize()

void Isis::AutoReg::SetSurfaceModelWindowSize ( int size )

inherited

Set the surface model window size.

The pixels in this window will be used to fit a surface model in order to compute sub-pixel accuracy. In some cases the default (3x3) and produces erroneous sub-pixel accuracy values.

If this method is not called, the window size defaults to 5 in the AutoReg object constructor.

Parameters

size	The size of the window must be three or greater and odd.

Exceptions

iException::User - "Invalid value for SurfaceModel WindowSize."

References _FILEINFO_, and Isis::IException::User.

Referenced by Isis::AutoReg::AutoReg(), and Isis::AutoReg::Parse().

◆ SetTolerance()

void Isis::AutoReg::SetTolerance ( double tolerance )

inherited

Set the tolerance for an acceptable goodness of fit.

If this method is not called, the tolerance value defaults to Isis::Null in the AutoReg object constructor.

Parameters

tolerance This tolerance is used to test against the goodness of fit returned by the MatchAlgorith method after a surface model has been fit. See TestGoodnessOfFit

Referenced by Isis::AutoReg::AutoReg(), and Isis::AutoReg::Parse().

◆ SubPixelAccuracy()

bool Isis::AutoReg::SubPixelAccuracy ( )

inlineinherited

Return whether this object will attempt to register to whole or sub-pixel accuracy.

Returns: on Is sub-pixel accuracy enabled?

Referenced by Isis::AutoReg::UpdatedTemplate().

◆ SubsearchValidPercent()

double Isis::AutoReg::SubsearchValidPercent ( ) const

inlineinherited

Return subsearch chip valid percent.

Referenced by Isis::AutoReg::UpdatedTemplate().

◆ Success()

bool Isis::AutoReg::Success ( ) const

inlineinherited

Return whether the match algorithm succeeded or not.

References Isis::AutoReg::SuccessPixel, and Isis::AutoReg::SuccessSubPixel.

Referenced by Isis::AutoReg::Register().

◆ Tolerance()

double Isis::AutoReg::Tolerance ( ) const

inlineinherited

Return match algorithm tolerance.

Referenced by Isis::AutoReg::Registration(), and Isis::AutoReg::UpdatedTemplate().

◆ UpdatedTemplate()

PvlGroup Isis::AutoReg::UpdatedTemplate ( )

inherited

Returns a PvlGroup containing the PvlKeywords of the parameters this object was most recently run with.

Because of publically accessible mutators, AutoReg's runtime parameters can change, and this version of the template is designed to reflect the paramters it was actually run with, as opposed to simply initialized with. If no mutators were ever called externally, then this PvlGroup should be the same as the one received from RegTemplate().

Returns: PvlGroup The keywords this object was most recently run with

References Isis::AutoReg::AlgorithmName(), Isis::AutoReg::DistanceTolerance(), Isis::AutoReg::GradientFilterString(), Isis::AutoReg::MinimumZScore(), Isis::AutoReg::PatternChip(), Isis::AutoReg::PatternValidPercent(), Isis::AutoReg::ReductionFactor(), Isis::AutoReg::SearchChip(), Isis::AutoReg::SubPixelAccuracy(), Isis::AutoReg::SubsearchValidPercent(), Isis::AutoReg::Tolerance(), Isis::toString(), and Isis::AutoReg::WindowSize().

◆ WindowSize()

double Isis::AutoReg::WindowSize ( ) const

inlineinherited

Return window size.

Referenced by Isis::AutoReg::UpdatedTemplate().

◆ WriteSubsearchChips()

void Isis::Gruen::WriteSubsearchChips ( const QString & pattern = "SubChip" )

Set up for writing subsearch for a a given registration call.

This method is provided to request the write of the subsearch chip at each iteration. This must be invoked prior to every call to AutoReg::Register() method. It will only write subchips from the Register() interface as it interates to a solution. Direct calls to Gruen methods don't iterate in the same fashion.

The "pattern" parameter is optional but is provided to direct the location and naming convention of each subsearch chip. The format for the output file name for each subsearch chip is comprised of the pattern parameter, call number, which can be retrieved by the CallCount() method and pertains to the call after the Register() method is invoked and the interation count. Below is a code example:

Gruen gruen(myPvldef);
 
// Set up pattern and search chips here
 
gruen.WriteSubsearchChips("/mydata/subchip");
gruen.Register();

Note that prior to each call to Register(), it must be called again in order for the subchips to be written. The last part of the parameter above, "subchip", cannot be a directory, but is a filename prefix. Assuming this is the first call to Register(), a series of cube subsearch chips will be written with the pattern "/work1/kbecker/subchipC000001IXXX.cub" where "C" indicates call count and the next 6 digits are the return of CallCount() method, "I" indicates the iteration count "XXX" of the algorithm() method. Note that the chip written for a particular iteration is what is provided as a parameter into the algorithm() method.

Parameters

pattern Specifies an optional directory and file pattern to write the subsearch chip at each algorithm iteration.

◆ ZScores()

void Isis::AutoReg::ZScores	(	double &	score1,
		double &	score2 ) const

inlineinherited

Return the ZScores of the pattern chip.

Parameters

score1	First Z Score
score2	Second Z Score

Member Data Documentation

◆ p_template

PvlObject Isis::AutoReg::p_template

protectedinherited

AutoRegistration object that created this projection.

Referenced by Isis::AutoReg::AutoReg(), and Isis::AutoReg::RegTemplate().

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

/Users/astamile/ISIS3/isis/src/base/objs/Gruen/Gruen.h
/Users/astamile/ISIS3/isis/src/base/objs/Gruen/Gruen.cpp

Public Types

Public Member Functions

Protected Member Functions

Protected Attributes

Detailed Description

Member Enumeration Documentation

◆ GradientFilterType

◆ RegisterStatus

Constructor & Destructor Documentation

◆ Gruen() [1/2]

◆ Gruen() [2/2]

◆ ~Gruen()

Member Function Documentation

◆ algorithm()

◆ AlgorithmName()

◆ AlgorithmStatistics()

◆ CallCount()

◆ ChipLine()

◆ ChipSample()

◆ CompareFits()

◆ CubeLine()

◆ CubeSample()

◆ Distance()

◆ DistanceTolerance()

◆ errorAnalysis()

◆ FitChip()

◆ getAffineConstraint()

◆ getAffineRadio()

◆ getAffineTolerance()

◆ getDefaultAffineRadio()

◆ getLastMatch()

◆ getSpiceConstraint()

◆ GoodnessOfFit()

◆ GradientFilterString()

◆ IdealFit()

◆ IsIdeal()

◆ MatchAlgorithm()

◆ MinimumZScore()

◆ MostLenientTolerance()

◆ Parse()

◆ PatternChip()

◆ PatternValidPercent()

◆ Reduce()

◆ ReducedFitChip()

◆ ReducedPatternChip()

◆ ReducedSearchChip()

◆ ReductionFactor()

◆ Register()

◆ Registration()

◆ RegistrationPatternChip()

◆ RegistrationSearchChip()

◆ RegistrationStatistics()

◆ RegTemplate()

◆ SearchChip()

◆ setAffineRadio() [1/2]

◆ setAffineRadio() [2/2]

◆ SetChipInterpolator()

◆ SetChipLine()

◆ SetChipSample()

◆ SetGoodnessOfFit()

◆ SetGradientFilterType()

◆ SetPatternValidPercent()

◆ SetPatternZScoreMinimum()

◆ SetReductionFactor()

◆ SetSubPixelAccuracy()

◆ SetSubpixelPosition()

◆ SetSubsearchValidPercent()

◆ SetSurfaceModelDistanceTolerance()

◆ SetSurfaceModelWindowSize()

◆ SetTolerance()

◆ SubPixelAccuracy()

◆ SubsearchValidPercent()

◆ Success()

◆ Tolerance()

◆ UpdatedTemplate()

◆ WindowSize()

◆ WriteSubsearchChips()

◆ ZScores()

Member Data Documentation

◆ p_template