Research On 3D Model Representation, Indexing And Retrieval Method Based On Data Mining

As the process of world globalization becomes more and more quickly, how to design new products quickly is much more vital than before. To another aspect, the development of computer techniques makes it easy that people can store CAD models in digital formats. User can view the models conveniently, and this makes it possible for CAD designers to reuse existing models. However, as the time goes, the number of models in the depository becomes large, and designers usually spend much more time on searching models which they want than before. How to efficiently retrieval models from the depository becomes important for the designers. To solve this problem, a new 3D CAD model retrieval method was proposed in this dissertation.The main contents of this dissertation are presented as below:(1) Use one of the CAD model descriptors, Median Axis Attribute adjacent Graph (MAAG) to describe the CAD model. One CAD model can be expressed well by its MAAG.(2) Extract frequent sub-patterns from the MAAG database of CAD model by data mining technique. This pattern set is used as the summary of the MAAG set.(3) Propose a new method to assess the similarity of two CAD model'MAAGs based on extracted subpattern set. The compute of maximum common subgraph can be avoided by this method.(4) A new index structure based on sub-pattern set is put forward in this dissertation. This new structure, usually named as Extended-Index-Tree, utilizes the inclusion relation between elements of the sub-pattern set to support fast-search operation of sub-pattern. With the help of this Extended-Index-Tree, the overall performance of the system is improved.(5) A prototype named Search Engine is implemented. Experiments show that the method proposed in this dissertation is validating.