Research On Surface Reconstruction Method Of 3D Discrete Point Cloud Triangular Mesh

With the development of computer science and technology,computer graphics &image and computer vision have penetrated into face recognition,driverless and other research areas.Computer-based three-dimensional modeling technology is to use computer to build 3D model of object.As the core technology of 3D modeling,surface reconstruction has always been a hot and difficult topic.This technology has been widely used in fields of automatic driving,robotic fields,industrial manufacturing and testing,archaeological and three-dimensional city modeling.At present,there are some problems in the algorithms of surface reconstruction,such as poor effect of non-closed surface reconstruction,low efficiency and poor smoothness.In order to solve the above problems,the paper selects the parameters with basis to improve the efficiency of the algorithm,increases the constraints to improve the surface reconstruction algorithm,and then improves the effect of non closed surface reconstruction.Finally,according to the subdivision surface reconstruction,the smoothness of the model is improved.The main research contents and conclusions are as follows.(1)In order to solve the problem that the effect of non-closed surface reconstruction is not ideal in point cloud data surface reconstruction,this paper constructs a advancing-front surface reconstruction algorithm based on 3D Delaunay triangulation.This algorithm that takes an unoriented point cloud as input,first of all,starting with a seed triangle from the Delaunay tetrahedron result.According to the triangle perimeter and other appropriate triangle selection criteria,the surface with correct geometry and topology is gradually formed by adding the linked Delaunay triangle from the current boundaries one by one.Using a wide range of datasets to evaluated this method,experimental result show that it could identify the surface boundary and remove the spurious surface at the boundary.(2)In order to solve the problem of parameter dependence and difficulty in parameter selection in surface reconstruction algorithm,KD-Tree is introduced as query index structure and applied to the k-nearest neighbor search to determine parameter threshold.Its processing is: firstly,the k-neighborhood set of the target point is established,and the Euclidean distance from the target point to the neighborhood point is calculated.Then,the Euclidean distance is arranged according to the size,and the maximum value is obtained.Finally,the maximum distance and the average distance of the whole point cloud data are obtained by traversing the whole point cloud data,and the threshold value of the parameter is set accordingly.The experimental results show that this method can avoid the problem of wasting a lot of time due to the wrong selection of parameters,improve the efficiency of the algorithm and reduce the time cost.(3)The algorithm constructed in this paper is compared with greedy projection algorithm,Poisson algorithm and BPA algorithm from five aspects of the non-uniform point cloud,sharp feature point cloud,open surface,closed surface and refined surface reconstruction.The experimental results show that: the algorithm proposed in this paper does not need to calculate the normal,and the parameter selection is more reasonable.It is more ideal for the reconstruction of models with boundary,non-uniform sampling data and models with sharp features.(4)In order to solve the problems of poor smoothness of the initial mesh obtained after reconstruction and difficult to express the real shape of the surface due to the unsatisfactory original point cloud data in solid modeling.Firstly,the experiments are carried out according to the geometric and topological rules of subdivision algorithm,and then the comparative analysis of four classical subdivision algorithms is carried out from three aspects of approximation degree,subdivision surface quality and subdivision complexity.Finally,based on the improved algorithm,the loop subdivision algorithm is selected to carry out the example verification,and the surface reconstruction results with good smoothness are obtained.