public class ClusterVisualWords
extends java.lang.Object
TupleDesc_F64 which can be used to identify frequent features, a.k.a words.
Internally it uses ComputeClusters and simply extracts the inner array
from the tuple.| Constructor and Description |
|---|
ClusterVisualWords(org.ddogleg.clustering.ComputeClusters<double[]> computeClusters,
int featureDOF,
long randomSeed)
Constructor which configures the cluster finder.
|
| Modifier and Type | Method and Description |
|---|---|
void |
addReference(boofcv.struct.feature.TupleDesc_F64 feature)
Add a feature to the list.
|
org.ddogleg.clustering.AssignCluster<double[]> |
getAssignment()
Returns a transform from point to cluster.
|
void |
process(int numberOfWords)
Clusters the list of features into the specified number of words
|
public ClusterVisualWords(org.ddogleg.clustering.ComputeClusters<double[]> computeClusters,
int featureDOF,
long randomSeed)
computeClusters - Cluster finding algorithm.featureDOF - Number of elements in the featurerandomSeed - Seed for random number generatorpublic void addReference(boofcv.struct.feature.TupleDesc_F64 feature)
feature - image feature. Reference to inner array is saved.public void process(int numberOfWords)
numberOfWords - Number of words/clusters it should findpublic org.ddogleg.clustering.AssignCluster<double[]> getAssignment()