A Multi-level Method For Fast Computing Of All-pairs 3D Inner-Distances

With the development of multi-media technology,3D information technology has achieved a wide range of applications in many fields of daily life so far,such as 3D computer games,medical imaging research,3D print,research and application on virtual reality,research on molecular biology,computer-aided design.Moreover,with the development of Internet,cloud computing and big data technology,3D shape can be foreseen to gain more and wider applications.Thus it is very meaningful to research on 3D shapes.Extraction of shape feature is a prior work of research on 3D shapes,therefore we put forward to a novel algorithm focus on 3D inner-distances,namely a multi-level method for fast computing of 3D inner-distances,aimed at meeting the requirement of feature extracting in large scale of shape models.Firstly,we introduce some basic information about 3D shapes,including the generation of 3D shape models,the storage of shape model files,basic classification of 3D shape feature extracting and the common 3D shape feature descriptors.Then we introduce the main algorithm in two parts: in the first part,we put forward to a top-down adaptive-voxelization algorithm based on octree structure,which is the premise of constructing the hierarchical graph and aimed at optimizing the storage of voxelization process;in the second part,we introduce the detail of our multi-level algorithm.Finally the performance of the algorithm is verified through experiments and comparision.