Skeleton-based 3D Model Descriptor and Its Application in Non-rigid Shape Retrieval

In this paper, we propose a novel skeleton-based 3D model descriptor to describe the shape of a triangular surface mesh. The main idea is to match skeleton tree by comparing the skeleton tree structure with shape surface area distribution, which is achieved in three steps as follows. First, based on the extracted skeletons of a non-rigid shape and geodesic distance computation, the center point in skeleton is defined and detected. Second, a skeleton tree is constructed based on the relationship between the center point and other discrete points in the skeleton. Finally, a correspondence between the skeleton tree and the area distribution of the non-rigid shape surface is established, and the skeleton-based descriptor is obtained. The advantages of our descriptor are three-fold: (1) scale invariance of geodesic distance and area distribution, (2) low computational complexity of geodesic distance, and (3) automatic topology repair. Experimental results demonstrate effectiveness and accuracy of our method, which is applicable to non-rigid shape comparison and retrieval.