LMedSLineCorrespondenceAffineTransformation2DRobustEstimator.java
/*
* Copyright (C) 2015 Alberto Irurueta Carro (alberto@irurueta.com)
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package com.irurueta.geometry.estimators;
import com.irurueta.algebra.AlgebraException;
import com.irurueta.geometry.AffineTransformation2D;
import com.irurueta.geometry.CoincidentLinesException;
import com.irurueta.geometry.Line2D;
import com.irurueta.numerical.robust.LMedSRobustEstimator;
import com.irurueta.numerical.robust.LMedSRobustEstimatorListener;
import com.irurueta.numerical.robust.RobustEstimator;
import com.irurueta.numerical.robust.RobustEstimatorException;
import com.irurueta.numerical.robust.RobustEstimatorMethod;
import java.util.List;
/**
* Finds the best affine 2D transformation for provided collections of matched
* 2D lines using LMedS algorithm.
*/
public class LMedSLineCorrespondenceAffineTransformation2DRobustEstimator extends
LineCorrespondenceAffineTransformation2DRobustEstimator {
/**
* 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.
*/
public static final double DEFAULT_STOP_THRESHOLD = 1e-6;
/**
* Minimum allowed stop threshold value.
*/
public static final double MIN_STOP_THRESHOLD = 0.0;
/**
* 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.
*/
private double stopThreshold;
/**
* Constructor.
*/
public LMedSLineCorrespondenceAffineTransformation2DRobustEstimator() {
super();
stopThreshold = DEFAULT_STOP_THRESHOLD;
}
/**
* Constructor with lists of lines to be used to estimate an affine 2D
* transformation.
* Lines 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.
*
* @param inputLines list of input lines to be used to estimate an affine
* 2D transformation.
* @param outputLines list of output lines to be used to estimate an affine
* 2D transformation.
* @throws IllegalArgumentException if provided lists of lines don't have
* the same size or their size is smaller than MINIMUM_SIZE.
*/
public LMedSLineCorrespondenceAffineTransformation2DRobustEstimator(
final List<Line2D> inputLines, final List<Line2D> outputLines) {
super(inputLines, outputLines);
stopThreshold = DEFAULT_STOP_THRESHOLD;
}
/**
* Constructor.
*
* @param listener listener to be notified of events such as when estimation
* starts, ends or its progress significantly changes.
*/
public LMedSLineCorrespondenceAffineTransformation2DRobustEstimator(
final AffineTransformation2DRobustEstimatorListener listener) {
super(listener);
stopThreshold = DEFAULT_STOP_THRESHOLD;
}
/**
* Constructor with listener and lists of lines to be used to estimate an
* affine 2D transformation.
* Lines 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.
*
* @param listener listener to be notified of events such as when estimation
* starts, ends or its progress significantly changes.
* @param inputLines list of input lines to be used to estimate an affine
* 2D transformation.
* @param outputLines list of output lines to be used to estimate an affine
* 2D transformation.
* @throws IllegalArgumentException if provided lists of lines don't have
* the same size or their size is smaller than MINIMUM_SIZE.
*/
public LMedSLineCorrespondenceAffineTransformation2DRobustEstimator(
final AffineTransformation2DRobustEstimatorListener listener,
final List<Line2D> inputLines, final List<Line2D> outputLines) {
super(listener, inputLines, outputLines);
stopThreshold = DEFAULT_STOP_THRESHOLD;
}
/**
* 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.
*
* @return stop threshold to stop the algorithm prematurely when a certain
* accuracy has been reached.
*/
public double getStopThreshold() {
return 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.
* 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.
*
* @param stopThreshold stop threshold to stop the algorithm prematurely
* when a certain accuracy has been reached.
* @throws IllegalArgumentException if provided value is zero or negative.
* @throws LockedException if robust estimator is locked because an
* estimation is already in progress.
*/
public void setStopThreshold(final double stopThreshold) throws LockedException {
if (isLocked()) {
throw new LockedException();
}
if (stopThreshold <= MIN_STOP_THRESHOLD) {
throw new IllegalArgumentException();
}
this.stopThreshold = stopThreshold;
}
/**
* Estimates an affine 2D transformation using a robust estimator and
* the best set of matched 2D lines correspondences found using the robust
* estimator.
*
* @return an affine 2D transformation.
* @throws LockedException if robust estimator is locked because an
* estimation is already in progress.
* @throws NotReadyException if provided input data is not enough to start
* the estimation.
* @throws RobustEstimatorException if estimation fails for any reason
* (i.e. numerical instability, no solution available, etc).
*/
@SuppressWarnings("DuplicatedCode")
@Override
public AffineTransformation2D estimate() throws LockedException, NotReadyException, RobustEstimatorException {
if (isLocked()) {
throw new LockedException();
}
if (!isReady()) {
throw new NotReadyException();
}
final var innerEstimator = new LMedSRobustEstimator<>(
new LMedSRobustEstimatorListener<AffineTransformation2D>() {
// line to be reused when computing residuals
private final Line2D testLine = new Line2D();
@Override
public int getTotalSamples() {
return inputLines.size();
}
@Override
public int getSubsetSize() {
return AffineTransformation2DRobustEstimator.MINIMUM_SIZE;
}
@Override
public void estimatePreliminarSolutions(
final int[] samplesIndices, final List<AffineTransformation2D> solutions) {
final var inputLine1 = inputLines.get(samplesIndices[0]);
final var inputLine2 = inputLines.get(samplesIndices[1]);
final var inputLine3 = inputLines.get(samplesIndices[2]);
final var outputLine1 = outputLines.get(samplesIndices[0]);
final var outputLine2 = outputLines.get(samplesIndices[1]);
final var outputLine3 = outputLines.get(samplesIndices[2]);
try {
final var transformation = new AffineTransformation2D(inputLine1, inputLine2, inputLine3,
outputLine1, outputLine2, outputLine3);
solutions.add(transformation);
} catch (final CoincidentLinesException e) {
// if lines are coincident, no solution is added
}
}
@Override
public double computeResidual(final AffineTransformation2D currentEstimation, final int i) {
final var inputLine = inputLines.get(i);
final var outputLine = outputLines.get(i);
// transform input line and store result in mTestLine
try {
currentEstimation.transform(inputLine, testLine);
return getResidual(outputLine, testLine);
} catch (final AlgebraException e) {
// this happens when internal matrix of affine transformation
// cannot be reverse (i.e. transformation is not well-defined,
// numerical instabilities, etc.)
return Double.MAX_VALUE;
}
}
@Override
public boolean isReady() {
return LMedSLineCorrespondenceAffineTransformation2DRobustEstimator.this.isReady();
}
@Override
public void onEstimateStart(final RobustEstimator<AffineTransformation2D> estimator) {
if (mListener != null) {
mListener.onEstimateStart(
LMedSLineCorrespondenceAffineTransformation2DRobustEstimator.this);
}
}
@Override
public void onEstimateEnd(final RobustEstimator<AffineTransformation2D> estimator) {
if (mListener != null) {
mListener.onEstimateEnd(
LMedSLineCorrespondenceAffineTransformation2DRobustEstimator.this);
}
}
@Override
public void onEstimateNextIteration(
final RobustEstimator<AffineTransformation2D> estimator, final int iteration) {
if (mListener != null) {
mListener.onEstimateNextIteration(
LMedSLineCorrespondenceAffineTransformation2DRobustEstimator.this,
iteration);
}
}
@Override
public void onEstimateProgressChange(
final RobustEstimator<AffineTransformation2D> estimator, final float progress) {
if (mListener != null) {
mListener.onEstimateProgressChange(
LMedSLineCorrespondenceAffineTransformation2DRobustEstimator.this,
progress);
}
}
});
try {
locked = true;
inliersData = null;
innerEstimator.setConfidence(confidence);
innerEstimator.setMaxIterations(maxIterations);
innerEstimator.setProgressDelta(progressDelta);
innerEstimator.setStopThreshold(stopThreshold);
final var transformation = innerEstimator.estimate();
inliersData = innerEstimator.getInliersData();
return attemptRefine(transformation);
} catch (final com.irurueta.numerical.LockedException e) {
throw new LockedException(e);
} catch (final com.irurueta.numerical.NotReadyException e) {
throw new NotReadyException(e);
} finally {
locked = false;
}
}
/**
* Returns method being used for robust estimation.
*
* @return method being used for robust estimation.
*/
@Override
public RobustEstimatorMethod getMethod() {
return RobustEstimatorMethod.LMEDS;
}
/**
* 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.
*
* @return standard deviation used for refinement.
*/
@Override
protected double getRefinementStandardDeviation() {
final var inliersData = (LMedSRobustEstimator.LMedSInliersData) getInliersData();
// avoid setting a threshold too strict
final var threshold = inliersData.getEstimatedThreshold();
return Math.max(threshold, stopThreshold);
}
}