Balanced hierarchical face clustering algorithm on triangle meshes

In this paper we propose a new algorithm of face clustering on triangle meshes, we call it balanced hierarchical face clustering (BHFC for short). Given a triangle mesh model, the algorithm iteratively generates a binary tree of face clusters on it. Each cluster consists of a set of connected faces. At every iteration, two connected clusters are chosen according to these rules we define, and they are merged into a new cluster. For each iteration there are two steps. In the first step, a set of cluster pair candidates is determined. In the second step, a single pair in the candidate set is chosen to merge. After the iteration, a hierarchical structure of clusters is constructed. The hierarchical structure is balanced, which avoids extreme size clusters appearing in the clustered result.