3D Model Retrieval based on both the Topology and Shape Features

The 3D model retrieval methods accounting for only global shape features are unable to compare models elaborately at a local level of detail.In this paper we present a new framework for 3D model retrieval considering both the global topology feature and the local shape feature of the candidate models.Firstly, we employ the Potential Field Method to acquire the skeleton of the model, and decompose it into several sub-parts according to the skeleton nodes.Then, we extract the local shape feature of each sub-part using the Spherical Harmonical Method.During the process of model matching, a three-step strategy is adopted.In the first step, we evaluate the similarity of global topology feature between the two skeletons and establish the corresponding relationship of the sub-parts between the two models.In the second step, we evaluate the similarity of local shape feature between each pair of corresponding sub-parts.The overall similarity of the two models is finally estimated by the weighted sum of the similarities on the above two aspects.Experimental results show that the result of this method is more satisfactory than those by the traditional methods based solely on the shape feature or the topology feature.Besides, this method supports the matching based on the local feature.