Package com.irurueta.geometry.estimators

Class LMedSPointCorrespondenceAffineTransformation2DRobustEstimator

java.lang.Object

com.irurueta.geometry.estimators.AffineTransformation2DRobustEstimator

com.irurueta.geometry.estimators.PointCorrespondenceAffineTransformation2DRobustEstimator

com.irurueta.geometry.estimators.LMedSPointCorrespondenceAffineTransformation2DRobustEstimator

public class LMedSPointCorrespondenceAffineTransformation2DRobustEstimator
extends PointCorrespondenceAffineTransformation2DRobustEstimator

Finds the best affine 2D transformation for provided collections of matched 2D points using LMedS algorithm.

Field Summary

Fields

Modifier and Type	Field	Description
static final double	DEFAULT_STOP_THRESHOLD	Default value to be used for stop threshold.
static final double	MIN_STOP_THRESHOLD	Minimum allowed stop threshold value.
private double	stopThreshold	Threshold to be used to keep the algorithm iterating in case that best estimated threshold using median of residuals is not small enough.

Fields inherited from class com.irurueta.geometry.estimators.PointCorrespondenceAffineTransformation2DRobustEstimator

DEFAULT_ROBUST_METHOD, inputPoints, outputPoints

Fields inherited from class com.irurueta.geometry.estimators.AffineTransformation2DRobustEstimator

confidence, covariance, DEFAULT_CONFIDENCE, DEFAULT_KEEP_COVARIANCE, DEFAULT_MAX_ITERATIONS, DEFAULT_PROGRESS_DELTA, DEFAULT_REFINE_RESULT, inliersData, keepCovariance, locked, MAX_CONFIDENCE, MAX_PROGRESS_DELTA, maxIterations, MIN_CONFIDENCE, MIN_ITERATIONS, MIN_PROGRESS_DELTA, MINIMUM_SIZE, mListener, progressDelta, refineResult

Constructor Summary

Constructors

Constructor	Description
LMedSPointCorrespondenceAffineTransformation2DRobustEstimator()	Constructor.
LMedSPointCorrespondenceAffineTransformation2DRobustEstimator(AffineTransformation2DRobustEstimatorListener listener)	Constructor.
LMedSPointCorrespondenceAffineTransformation2DRobustEstimator(AffineTransformation2DRobustEstimatorListener listener, List<Point2D> inputPoints, List<Point2D> outputPoints)	Constructor with listener and lists of points to be used to estimate an affine 2D transformation.
LMedSPointCorrespondenceAffineTransformation2DRobustEstimator(List<Point2D> inputPoints, List<Point2D> outputPoints)	Constructor with lists of points to be used to estimate an affine 2D transformation.

Method Summary

Modifier and Type	Method	Description
AffineTransformation2D	estimate()	Estimates an affine 2D transformation using a robust estimator and the best set of matched 2D point correspondences found using the robust estimator.
com.irurueta.numerical.robust.RobustEstimatorMethod	getMethod()	Returns method being used for robust estimation.
protected double	getRefinementStandardDeviation()	Gets standard deviation used for Levenberg-Marquardt fitting during refinement.
double	getStopThreshold()	Returns threshold to be used to keep the algorithm iterating in case that best estimated threshold using median of residuals is not small enough.
void	setStopThreshold(double stopThreshold)	Sets threshold to be used to keep the algorithm iterating in case that best estimated threshold using median of residuals is not small enough.

Methods inherited from class com.irurueta.geometry.estimators.PointCorrespondenceAffineTransformation2DRobustEstimator

attemptRefine, create, create, create, create, create, create, create, create, create, create, create, create, create, create, create, create, getInputPoints, getOutputPoints, getQualityScores, isReady, setPoints, setQualityScores

Methods inherited from class com.irurueta.geometry.estimators.AffineTransformation2DRobustEstimator

createFromLines, createFromLines, createFromLines, createFromLines, createFromLines, createFromLines, createFromLines, createFromLines, createFromPoints, createFromPoints, createFromPoints, createFromPoints, createFromPoints, createFromPoints, createFromPoints, createFromPoints, getConfidence, getCovariance, getInliersData, getListener, getMaxIterations, getProgressDelta, isCovarianceKept, isListenerAvailable, isLocked, isResultRefined, setConfidence, setCovarianceKept, setListener, setMaxIterations, setProgressDelta, setResultRefined

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Details

DEFAULT_STOP_THRESHOLD

public static final double DEFAULT_STOP_THRESHOLD

Default value to be used for stop threshold. Stop threshold can be used to keep the algorithm iterating in case that best estimated threshold using median of residuals is not small enough. Once a solution is found that generates a threshold below this value, the algorithm will stop. The stop threshold can be used to prevent the LMedS algorithm iterating too many times in cases where samples have a very similar accuracy. For instance, in cases where proportion of outliers is very small (close to 0%), and samples are very accurate (i.e. 1e-6), the algorithm would iterate for a long time trying to find the best solution when indeed there is no need to do that if a reasonable threshold has already been reached. Because of this behaviour the stop threshold can be set to a value much lower than the one typically used in RANSAC, and yet the algorithm could still produce even smaller thresholds in estimated results.

See Also:

• Constant Field Values

MIN_STOP_THRESHOLD

public static final double MIN_STOP_THRESHOLD

Minimum allowed stop threshold value.

See Also:

• Constant Field Values

stopThreshold

private double stopThreshold

Threshold to be used to keep the algorithm iterating in case that best estimated threshold using median of residuals is not small enough. Once a solution is found that generates a threshold below this value, the algorithm will stop. The stop threshold can be used to prevent the LMedS algorithm iterating too many times in cases where samples have a very similar accuracy. For instance, in cases where proportion of outliers is very small (close to 0%), and samples are very accurate (i.e. 1e-6), the algorithm would iterate for a long time trying to find the best solution when indeed there is no need to do that if a reasonable threshold has already been reached. Because of this behaviour the stop threshold can be set to a value much lower than the one typically used in RANSAC, and yet the algorithm could still produce even smaller thresholds in estimated results.

Constructor Details

LMedSPointCorrespondenceAffineTransformation2DRobustEstimator

public LMedSPointCorrespondenceAffineTransformation2DRobustEstimator()

Constructor.

LMedSPointCorrespondenceAffineTransformation2DRobustEstimator

public LMedSPointCorrespondenceAffineTransformation2DRobustEstimator(List<Point2D> inputPoints,
 List<Point2D> outputPoints)

Constructor with lists of points to be used to estimate an affine 2D transformation. Points in the list located at the same position are considered to be matched. Hence, both lists must have the same size, and their size must be greater or equal than MINIMUM_SIZE.

Parameters:

inputPoints - list of input points to be used to estimate an affine 2D transformation.

outputPoints - list of output points to be used to estimate an affine 2D transformation.

Throws:

IllegalArgumentException - if provided lists of points don't have the same size or their size is smaller than MINIMUM_SIZE.

LMedSPointCorrespondenceAffineTransformation2DRobustEstimator

public LMedSPointCorrespondenceAffineTransformation2DRobustEstimator(AffineTransformation2DRobustEstimatorListener listener)

Constructor.

Parameters:

listener - listener to be notified of events such as when estimation starts, ends or its progress significantly changes.

LMedSPointCorrespondenceAffineTransformation2DRobustEstimator

public LMedSPointCorrespondenceAffineTransformation2DRobustEstimator(AffineTransformation2DRobustEstimatorListener listener,
 List<Point2D> inputPoints,
 List<Point2D> outputPoints)

Constructor with listener and lists of points to be used to estimate an affine 2D transformation. Points in the list located at the same position are considered to be matched. Hence, both lists must have the same size, and their size must be greater or equal than MINIMUM_SIZE.

Parameters:

listener - listener to be notified of events such as when estimation starts, ends or its progress significantly changes.

inputPoints - list of input points to be used to estimate an affine 2D transformation.

outputPoints - list of output points to be used to estimate an affine 2D transformation.

Throws:

IllegalArgumentException - if provided lists of points don't have the same size or their size is smaller than MINIMUM_SIZE.

Method Details

getStopThreshold

public double getStopThreshold()

Returns threshold to be used to keep the algorithm iterating in case that best estimated threshold using median of residuals is not small enough. Once a solution is found that generates a threshold below this value, the algorithm will stop. The stop threshold can be used to prevent the LMedS algorithm iterating too many times in cases where samples have a very similar accuracy. For instance, in cases where proportion of outliers is very small (close to 0%), and samples are very accurate (i.e. 1e-6), the algorithm would iterate for a long time trying to find the best solution when indeed there is no need to do that if a reasonable threshold has already been reached. Because of this behaviour the stop threshold can be set to a value much lower than the one typically used in RANSAC, and yet the algorithm could still produce even smaller thresholds in estimated results.

Returns:

stop threshold to stop the algorithm prematurely when a certain accuracy has been reached.

setStopThreshold

public void setStopThreshold(double stopThreshold)
                      throws LockedException

Sets threshold to be used to keep the algorithm iterating in case that best estimated threshold using median of residuals is not small enough. Once a solution is found that generates a threshold below this value, the algorithm will stop. The stop threshold can be used to prevent the LMedS algorithm iterating too many times in cases where samples have a very similar accuracy. For instance, in cases where proportion of outliers is very small (close to 0%), and samples are very accurate (i.e. 1e-6), the algorithm would iterate for a long time trying to find the best solution when indeed there is no need to do that if a reasonable threshold has already been reached. Because of this behaviour the stop threshold can be set to a value much lower than the one typically used in RANSAC, and yet the algorithm could still produce even smaller thresholds in estimated results.

Parameters:

stopThreshold - stop threshold to stop the algorithm prematurely when a certain accuracy has been reached.

Throws:

IllegalArgumentException - if provided value is zero or negative.

LockedException - if robust estimator is locked because an estimation is already in progress.

estimate

public AffineTransformation2D estimate()
                                throws LockedException,
NotReadyException,
com.irurueta.numerical.robust.RobustEstimatorException

Estimates an affine 2D transformation using a robust estimator and the best set of matched 2D point correspondences found using the robust estimator.

Specified by:

estimate in class AffineTransformation2DRobustEstimator

Returns:

an affine 2D transformation.

Throws:

LockedException - if robust estimator is locked because an estimation is already in progress.

NotReadyException - if provided input data is not enough to start the estimation.

com.irurueta.numerical.robust.RobustEstimatorException - if estimation fails for any reason (i.e. numerical instability, no solution available, etc).

getMethod

public com.irurueta.numerical.robust.RobustEstimatorMethod getMethod()

Returns method being used for robust estimation.

Specified by:

getMethod in class AffineTransformation2DRobustEstimator

Returns:

method being used for robust estimation.

getRefinementStandardDeviation

protected double getRefinementStandardDeviation()

Gets standard deviation used for Levenberg-Marquardt fitting during refinement. Returned value gives an indication of how much variance each residual has. Typically, this value is related to the threshold used on each robust estimation, since residuals of found inliers are within the range of such threshold.

Specified by:

getRefinementStandardDeviation in class AffineTransformation2DRobustEstimator

Returns:

standard deviation used for refinement.