boofcv.factory.feature.associate

Class FactoryAssociation

java.lang.Object

boofcv.factory.feature.associate.FactoryAssociation

public class FactoryAssociation
extends java.lang.Object

Creates algorithms for associating TupleDesc_F64 features.

Constructor Summary

Constructors

Constructor and Description
FactoryAssociation()

Method Summary

Modifier and Type	Method and Description
static <D> ScoreAssociation<D>	defaultScore(java.lang.Class<D> tupleType) Given a feature descriptor type it returns a "reasonable" default ScoreAssociation.
static <D> AssociateDescription<D>	greedy(ScoreAssociation<D> score, double maxError, boolean backwardsValidation) Returns an algorithm for associating features together which uses a brute force greedy algorithm.
static AssociateDescription<TupleDesc_F64>	kdRandomForest(int dimension, int maxNodesSearched, int numTrees, int numConsiderSplit, long randomSeed) Approximate association using multiple random K-D trees (random forest) for descriptors with a high degree of freedom, e.g.
static AssociateDescription<TupleDesc_F64>	kdtree(int dimension, int maxNodesSearched) Approximate association using a K-D tree degree of moderate size (10-15) that uses a best-bin-first search order.
static <D> ScoreAssociation<D>	scoreEuclidean(java.lang.Class<D> tupleType, boolean squared) Scores features based on the Euclidean distance between them.
static <D> ScoreAssociation<D>	scoreHamming(java.lang.Class<D> tupleType) Hamming distance between two binary descriptors.
static ScoreAssociation<NccFeature>	scoreNcc() Scores features based on their Normalized Cross-Correlation (NCC).
static <D> ScoreAssociation<D>	scoreSad(java.lang.Class<D> tupleType) Scores features based on Sum of Absolute Difference (SAD).

Methods inherited from class java.lang.Object

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

Constructor Detail

FactoryAssociation

public FactoryAssociation()

Method Detail

greedy

public static <D> AssociateDescription<D> greedy(ScoreAssociation<D> score,
                                                 double maxError,
                                                 boolean backwardsValidation)

Returns an algorithm for associating features together which uses a brute force greedy algorithm. See AssociateGreedy for details.

Type Parameters:

D - Data structure being associated

Parameters:

score - Computes the fit score between two features.

maxError - Maximum allowed error/fit score between two features. To disable set to Double.MAX_VALUE

backwardsValidation - If true associations are validated by associating in the reverse direction. If the forward and reverse matches fit an association is excepted.

Returns:

AssociateDescription

kdtree

public static AssociateDescription<TupleDesc_F64> kdtree(int dimension,
                                                         int maxNodesSearched)

Approximate association using a K-D tree degree of moderate size (10-15) that uses a best-bin-first search order.

Parameters:

dimension - Number of elements in the feature vector

maxNodesSearched - Maximum number of nodes it will search. Controls speed and accuracy.

Returns:

Association using approximate nearest neighbor

See Also:

AssociateNearestNeighbor, KdTreeSearch1Bbf

kdRandomForest

public static AssociateDescription<TupleDesc_F64> kdRandomForest(int dimension,
                                                                 int maxNodesSearched,
                                                                 int numTrees,
                                                                 int numConsiderSplit,
                                                                 long randomSeed)

Approximate association using multiple random K-D trees (random forest) for descriptors with a high degree of freedom, e.g. > 20

Parameters:

dimension - Number of elements in the feature vector

maxNodesSearched - Maximum number of nodes it will search. Controls speed and accuracy.

numTrees - Number of trees that are considered. Try 10 and tune.

numConsiderSplit - Number of nodes that are considered when generating a tree. Must be less than the point's dimension. Try 5

randomSeed - Seed used by random number generator

Returns:

Association using approximate nearest neighbor

See Also:

AssociateNearestNeighbor, KdForestBbfSearch

defaultScore

public static <D> ScoreAssociation<D> defaultScore(java.lang.Class<D> tupleType)

Given a feature descriptor type it returns a "reasonable" default ScoreAssociation.

Parameters:

tupleType - Class type which extends TupleDesc

Returns:

A class which can score two potential associations

scoreSad

public static <D> ScoreAssociation<D> scoreSad(java.lang.Class<D> tupleType)

Scores features based on Sum of Absolute Difference (SAD).

Parameters:

tupleType - Type of descriptor being scored

Returns:

SAD scorer

scoreNcc

public static ScoreAssociation<NccFeature> scoreNcc()

Scores features based on their Normalized Cross-Correlation (NCC).

Returns:

NCC score

scoreEuclidean

public static <D> ScoreAssociation<D> scoreEuclidean(java.lang.Class<D> tupleType,
                                                     boolean squared)

Scores features based on the Euclidean distance between them. The square is often used instead of the Euclidean distance since it is much faster to compute.

Parameters:

tupleType - Type of descriptor being scored

squared - IF true the distance squared is returned. Usually true

Returns:

Euclidean distance measure

scoreHamming

public static <D> ScoreAssociation<D> scoreHamming(java.lang.Class<D> tupleType)

Hamming distance between two binary descriptors.

Parameters:

tupleType - Type of descriptor being scored

Returns:

Hamming distance measure