Research On A 3D Model Shape Descriptor Based On Topology Analysis And Its Applications

Topological features of 3D models are relatively high-level representations and close to human intuitive perception, which makes them useful for defining a more natural 3D representation. Most existing algorithms are restricted to the basic topological properties such as connectivity which makes their representations not accurate enough.Our work is based on topology analysis and discusses a novel 3D model shape descriptor, Algebraic Topology-based Segmentation Graph (ATSegGraph). By using simplicial complex growing to extract the minimal loop constraints, we obtain the fusion of topological and geometric features. This breaks through the restrictions of the existing topological features, and potentially can simplify 3D algorithms and represent models in a more accurate way.Based on the descriptor, we further discuss its typical 3D applications of mesh segmentation and retrieval.Automatic segmentation of 3D mesh models is one of the recent interests in computer graphics and plays an important role in many applications. By using the ATSegGraph, our method can successfully segment given 3D mesh models without any predefined parameters. Comparison results with other semi-automatic segmentation methods show that our algorithm accomplishes automatic 3D segmentation effectively.With the recent increase in the number of 3D models, development of the technology for effective retrieval of 3D models has become an important issue. Experimental results show that we can calculate the similarity between different models and make preparations for retrieval works by using ATSegGraph and defining appropriate similarity metric.