The Research Of 3D Model Correspondence Analysis Based On Diffusion Geometry

Three-dimensional model as an emerging digital geometric media,more and more integrate into people's daily life.For example,in the 3D printing,industrial establishment,3D game animation,film and television special effects,biological medicine and other technology fields have been practical application.At the same time,with the people of the constant pursuit of high technology,the number and complexity of three-dimensional models grows exponentially.Traditional geometric processing techniques only use the local geometric properties of the model,in the face of such a complex model processing tasks are often difficult to get the expected effect.With the further innovation of three-dimensional model geometry processing technology,researchers have come to realize and begin to understand,analyze and deal with the structural information of model.A series of shape analysis methods are proposed,which are oriented to semantic structure of the model.In recent years,the structural shape analysis of 3D model has become a hot topic in the field of computer graphics,digital geometry processing and so on.This paper analyzes the related research,aiming at the limitations of the current research,from the two aspects of the model itself and the between models to analyze the intrinsic structural similarity of the three-dimensional model,and propose a theoretical framework for shape analysis based on diffusion distance and geometric constraint.The main work of this paper is as follows:1.Introduces the main research contents of structural analysis of 3D model.2.Introduces the basic theory of this paper.The experiment analyzes the advantages and characteristics of the application of the diffusion geometry theory to the non-rigid 3D models correspondence analysis.3.Analyze the shape correspondence of the model from a single model.Using the difference of the heat diffusion distance between vertices of the model,so that construct the feature similarity matrix of the model,and fuse to the geometric constraint of the vertex neighborhood relation.Finally,the self-correspondence of the model is obtained,which can effectively detect he intrinsic symmetry of the 3D models.4.Fusion to the heat kernel signature and more robust interior space volume features,from the aspects of the different model to analysis the similarity of the models.This method can effectively reveal the unified structural characteristics of similar models,and improve the efficiency of shape matching and topology stability.