Cluster

Parameters for clustering selected hypotheses. The clustering allows PartFinder to

• distribute selected hypotheses more uniformly over a 3D scene,

• select fewer hypotheses thus spend less time for their refinement,

• utilize a priori knowledge about the number of objects on the 3D scene.

The clustering algorithm is hierarchical: a cluster is formed around a seed that is a hypothesis with the highest 2D score and does not belong to any cluster already found. Such cluster includes other free hypotheses which are in the vicinity of the seed.

Format

Object

Please refer to the sub node documentation.