Markov Models And Hidden Markov Model-based Three-dimensional Model Of Classification Research

With rapid development of the three-dimensional (3D) modeling technology and Internet technologies, the amount of 3D models take on an rapid increasing trend. And 3D models play a very important role in many fields, such as molecular biology, cultural heritage protection, computer-aided design and manufacturing and so on. This paper proposes a new method of 3D object classification based on existing statistical feature extraction method and Markov model to arrive accuracy and efficiently classification for the mass of 3D models. The main work and progress are listed as follows:(1) Summary and analysis of the 3D model retrieval and Classification's status and related technologies;(2) Introduce the existing feature extraction algorithms of 3D models, they include the visual feature extraction method, the geometric feature extraction method, the axial skeleton feature extraction method and the statistical feature extraction method. Then we analysis their advantages and disadvantages.And mainly introduce three types of the statistical feature extraction method:the shape histogram algorithm, point density algorithm, the weighted point set algorithm.(3) Propose a new method of 3D object classification based on Markov model through understanding the Markov process and the Markov chain's basic concepts and their classification. This algorithm uses Markov model's advantages in dealing with close relationship, through segmenting the 3D model and extracting features from the model to building the Markov model. This algorithm model takes the local characteristics and their closely relationship into consideration, rather than the traditional 3D model retrival algorithm which only match the corresponding local statistical features. At last we use the part of the Princeton University's 3D model database to do a experiment Princeton, the experimental results prove that the algorithm has a very good classification results and the factors that affect the classification results are less.(4) Introduce the basic concepts and three basic problems of the Hidden Markov Model(HMM), then three algorithms for the three basic problems are proved and propose a method of building a HMM for a object. At last we show that classification is also valid using HMM through a experiment for the simulation models, which provide a foundation for the further research of 3D model classification.