Class ClusterFinder

java.lang.Object
org.jlab.rec.dc.cluster.ClusterFinder

public class ClusterFinder extends Object
A hit pruning algorithm to reject noise that gives a pattern of hits that are continguous in the same layer The algorithm first puts the hits in arrays according to their layer and wire number. Each such array contains all the hits in the same layer. The algorithm then collects groups of contiguous hits into a list of hits. The n-first and n-last hits in the list are kept, and all other hits inbetween pruned. The value of n depends on the size of the list. A loose clustering algorithm loops over all superlayers, in a sector and finds groups of hits with contiguous wire index numbers. These clusters (called clumps of hits) are delimited by layers with no hits at a particular wire coordinate. These clusters are then refined using fits to their respective wire indexes as a function of layer number to identify parallel tracks or overlapping track candidates.