3D Mesh Segmentation Based On The Strength Of Boundary

In the computer graphics research Areas, three-dimensional mesh segmentation and retrieval technology has become a hotpot recently. With the development of three- dimensional scanning technology and three-dimensional modeling technology, 3D mesh model is increasingly used in 3D digital animation, virtual reality, realistic modeling of the geometry processing and other research areas. Gradually, it replaces the traditional surface model, become the mainstream model of the digital geometric modeling. Thus, the research to the grid model began to refine all aspects of treatment. Mesh segmentation and retrieval which is the highly representative of the research.3D mesh segmentation can help in many aspects, such as parameterization, shape matching, texture mapping, multi-resolution modeling, mesh editing, deformation, compression, animation and so on. In recent years, many excellent mesh segmentation algorithms appeared. There are Shape Diameter, Normalized Cuts, Core Extraction, Randomized Cuts methods, and so on. Each method has its own applicable model, so there are no single segmentation evaluation criteria to determine the merits of various algorithms. Thus, 3D mesh model segmentation evaluation techniques will also develop. 3D mesh retrieval is one of the applications of mesh segmentation, the main design goals of which are how to get similar shape characteristics model from large model base. Thus reuse and manage of the existing grid more effective.This paper presents a fast segmentation algorithm based on the grid boundary geometry information of triangular meshes. At first, according to the topological relation of original mesh faces, the algorithm sets dual graph and default weights for the mesh faces with geometry information. Use the k-way multilevel partitioning methods to divide the dual graph obtaining pre-segmentation areas and initial boundary for each area, then search the characteristic contour of the module to determine the characteristic boundary. The boundary strength is defined as a deformation model of the divided areas'boundary. By minimizing the energy function of the deformation model to promote the initial boundary approach to the characteristic boundary. Eventually, split the original model into several meaningful sub-divided parts in accordance with the minima rule. The experimental results show that our algorithm is efficient and effective, especially for a variety of triangular mesh models with sharp feature. For the model retrieve of large-scale three-dimensional inefficiency, this paper presents a three-dimensional model retrieval method based on the idea of incremental clustering. Firstly, for the models among the model base, construct a retrieval words codebook. Then extract the feature points to attain the feature vector of models according to the feature histogram, after that, followed by an incremental clustering method and update retrieval words codebook. Finally, a feature vector matching method is used to determine whether the model base contain the models which is relation to the target model. Experimental results show that our implementation can get the retrieval result rapidly and precisely.