Package net.preibisch.mvrecon.process.pointcloud.pointdescriptor.matcher

Class SubsetMatcher

java.lang.Object

net.preibisch.mvrecon.process.pointcloud.pointdescriptor.matcher.SubsetMatcher

All Implemented Interfaces:

Matcher

Direct Known Subclasses:

ModelPriorSubsetMatcher

public class SubsetMatcher
extends Object
implements Matcher

Constructor Summary

Constructors

Constructor	Description
SubsetMatcher(int subsetSize, int numNeighbors)	Matches n out of m neighbor points, the effort increases exponentially(!)

Method Summary

Modifier and Type	Method	Description
static int[][]	computePD(int n, int k, int offset)
protected static int[][]	computePDRecursive(int tolerance, int n, int offset)	Computes recursively how to create different PointDescriptors of the same Point when some tolerance is allowed, e.g. there are outliers which should be identified.
ArrayList<ArrayList<mpicbg.models.PointMatch>>	createCandidates(AbstractPointDescriptor<?,?> pd1, AbstractPointDescriptor<?,?> pd2)
int[][]	getNeighbors()
double	getNormalizationFactor(ArrayList<mpicbg.models.PointMatch> matches, Object fitResult)	Computes a normalization factor for the case that the different set of PointMatches are not comparable (for example number of neighbors used is not constant)
int	getNumCombinations()
int	getNumMatchings()
int	getNumNeighbors()
int	getRequiredNumNeighbors()
int	getSubsetSize()

Methods inherited from class java.lang.Object

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

Constructor Details

SubsetMatcher

public SubsetMatcher(int subsetSize,
 int numNeighbors)

Matches n out of m neighbor points, the effort increases exponentially(!) For example SubsetMatcher(4,4) is identical to calling SimpleMatcher(4)

Parameters:

subsetSize - (n) - how many neighbors are matched in each try

numNeighbors - (m) - out of how many neighbors to choose

Method Details

getNeighbors

public int[][] getNeighbors()

getSubsetSize

public int getSubsetSize()

getNumNeighbors

public int getNumNeighbors()

getNumCombinations

public int getNumCombinations()

getNumMatchings

public int getNumMatchings()

getRequiredNumNeighbors

public int getRequiredNumNeighbors()

Specified by:

getRequiredNumNeighbors in interface Matcher

Returns:

The number of nearest neighbors required for this Matcher

createCandidates

public ArrayList<ArrayList<mpicbg.models.PointMatch>> createCandidates(AbstractPointDescriptor<?,?> pd1,
 AbstractPointDescriptor<?,?> pd2)

Specified by:

createCandidates in interface Matcher

Parameters:

pd1 - - pointdescriptor 1

pd2 - - pointdescriptor 2

Returns:

An ArrayList of corresponding set of PointMatches which contain LinkedPoints linking to the actual Point instance they are created from

getNormalizationFactor

public double getNormalizationFactor(ArrayList<mpicbg.models.PointMatch> matches,
 Object fitResult)

Description copied from interface: Matcher

Computes a normalization factor for the case that the different set of PointMatches are not comparable (for example number of neighbors used is not constant)

Specified by:

getNormalizationFactor in interface Matcher

Parameters:

matches - the set of PointMatches

fitResult - - potential data needed to compute it

Returns:

The normalization factor for a certain set of PointMatches

computePD

public static int[][] computePD(int n,
 int k,
 int offset)

computePDRecursive

protected static int[][] computePDRecursive(int tolerance,
 int n,
 int offset)

Computes recursively how to create different PointDescriptors of the same Point when some tolerance is allowed, e.g. there are outliers which should be identified. For 3 neighbors allowing 2 outliers and starting with the second nearest neigbor (the nearest one is in most cases the one you use as input) the result looks like that: (Each coloum will be one PointDescriptor) |1| |1| |1| |2| |1| |1| |2| |1| |2| |3| |2| |2| |3| |3| |2| |3| |3| |4| |4| |4| |3| |4| |4| |4| |5| |5| |5| |5| |5| |5|

Parameters:

tolerance - - How many tolerance is accepted [0 ... m]

n - - initialized with the number of neighbors

offset - - the starting position for the neighbors, usually 1

Returns:

an array containing the neighbors for each PointDescriptor as array