Topological Decompositions for 3D Non-manifold Simplicial Shapes

Modeling and understanding complex non-manifold shapes is a key issue in several applications including form-feature identification in CAD/CAE, and shape recognition for Web searching. Geometric shapes are commonly discretized as simplicial 2- or 3-complexes embedded in the 3D Euclidean space. The topological structure of a non-manifold simplicial shape can be analyzed through its decomposition into a collection of components with simpler topology. The granularity of the decomposition depends on the combinatorial complexity of the components. In this paper, we present topological tools for structural analysis of three-dimensional non-manifold shapes. This analysis is based on a topological decomposition at two different levels. We discuss the topological properties of the components at each level, and we present algorithms for computing such decompositions. We investigate the relations among the components, and propose a graph-based representation for such relations.