Research On Evaluation Method Of Integral And Local Similarity Of 3D Models

The method of from top to bottom,from coarse to fine,from whole to local is widely adopted in modern product research and development.However,in view of the fact that even if structure of a part is very simple,it still needs a considerable amount of time for function?structure?process and other aspects redesign.If the existing products or technologies through appropriate changes can meet the new functional requirements,then not only can we make sure the quality of products,but also effectively reduce costs and strengthen the inheritance relationship between components?assembly?products.Therefore,design reuse for enterprises is important in terms of strategic significance.Because of the rapid development of 3D modeling theory and graphics hardware technology,the number of 3D model libraries such as modern industrial design field,game field and virtual reality becomes ever-increasing.3D models have been widely used because they contain a multidimensional set of information and rich visual details.Product life cycle management based on three-dimensional digital models is the development trend of modern product design and manufacturing;at the same time,enterprises in the production process have accumulated a large number of 3D models and the question we are faced at is how to retrieve specific models according to needs from large database,so the information and knowledge contained in them can be fully reused.It is one of the most important ways to speed up the product development process and improve the quality of product.Compared with traditional mechanical products,design process of agricultural machine is more difficult because of complexity of interactive objects.Adopting more advanced design methods has irreplaceable significance to improvement of China's overall agricultural mechanization's level.In view of the fact that text matching methods based on keywords have been unable to adapt to the reuse requirements of models in new area,content-based 3D model automatic retrieval technology has broad application prospects.This paper includes three following research aspects:1.A models' overall similarity evaluation algorithm based on distance-area.The idea behind this is that adopting statistical characteristics of three-dimensional models,whose principle is easy to understand,always has good effect so that it has been widely researched.In this paper,we optimize random points and area ratio between models' surfaces and surfaces is added to describe models further.Finally,offline feature extraction technology and online model matching technology is adopted.Experiments show that resolution of complex models is improved and time performance during which is acceptable;2.A local structure mining method through simulated annealing,in view of the fact that most general evaluation methods are not suit for local structures.Based on model attribute graphs after summarizing geometric and topology properties of models,our algorithm can retrieve local structures by building corresponding adjacency matrixes and finding out maximum cliques from association graphs with the comparison of ant colony and genetic algorithm.Experimental results show that the algorithm can be used as the initial heuristic local structure mining method compared with other modern artificial algorithms;3.Aiming at the two algorithms,Visual Studio.NET2008 being used as the integrated development environment,Open Cascade as the geometric platform,the professional chassis models with general models as library,a 3D digital model retrieval system is developed to facilitate the design reuse.