Optimization Of Finite Element Model Based On Adaptive Poisson-Disk Sampling

Point Sampling is an important direction in computer graphics algorithms.In this paper,a new algorithm is proposed,which combines the elimination sampling method and the sampling domain decomposition algorithm,and achieves sampling maximization on the premise of good blue noise property of points set.The main contents of this paper are summarized as follows:(1)We propose an adaptive Poisson disk sampling algorithm based on metric driven and applied it to mesh optimization.At the beginning,the elimination algorithm is used to obtain the initial set of points,and then with the help of dart-throwing and subdivision of sampling region methods,the sampling point set is also maximized.In this paper,we first discussed the most widely used dart-throwing algorithm,which are well-known for its good blue noise property,and the elimination method with the best Poisson blue noise characteristics as we have known.After that,aiming at the shortcomings of the dart-throwing algorithm,the paper analyzed the existing optimization methods on efficiency and point set maximization,including background grid method,gap processing method of the not sampled area,etc.In consideration of the advantages and disadvantages of these methods,we get a new algorithm of planar domain sampling framework,which are inspired by the combination of dart-throwing method,background grid method and elimination method.The measurement is introduced to control points' distribution,adaptive sampling points set is obtained.In the experimental part,auxiliary algorithm is realized by using CGAL library.In the analysis of sampling point set is applied to the mesh reconstruction results,used by the quality index of the grid unit,including the angle of triangular element in each interval percentage and the degree of each grid vertex.Experiments show that this algorithm can be used as a good choice of mesh reconstruction task,achieved the desired results.(2)To generalize 2D sampling framework to mesh surface,get sampling points set of surface model,we specify metric attribute of sampling points according to its Gauss curvature,construct convenient points' neighborhood searching data structure and point-face index mechanism,improved corresponding mesh unit's quality.