3D Shape Inner-Distance Computing for Shape Matching Based on Adaptive Volume Representation

3D Inner-distance is defined as the length of the shortest path between two points in a 3D shape. Since this distance is robust to articulation, it can precisely reflect the distance between two points with large but volume-preserving shape deformation, which is important for 3D shape analysis fields, especially for non-rigid 3D shape deformations. In this paper, we propose an adpative volume representation for 3D shapes, which can drastically speed up the process of computing 3D inner distances. Furthermore, based on 3D inner distances, we construct a global shape feature descriptor. The experiments illustrate this feature descriptor outperform those based on Euclidean distances and Geodesic distances for shape matching.