Research On Point Cloud Resampling And Surface Reconstruction

Surface reconstruction of point cloud is to connect point cloud topologically and restore the surface shape of the model,which is one of the key technologies of reverse engineering.The recovery of the sharp features of the surface of the model requires reconstruction algorithm.In the process of model sampling,the number of point cloud will affect the time and precision of surface reconstruction.The resampling of point cloud can increase or decrease the number of point cloud to ensure the accuracy and efficiency of reconstruction.In this thesis,the resampling and surface reconstruction algorithms of point cloud are studied.For large-scale point clouds,surface reconstruction is time-consuming and takes up a lot of memory.In order to improve reconstruction efficiency,a point cloud simplification method is studied.Firstly,the point cloud is classified into feature points and non-feature points according to the surface variation of each point.Then,the space of point cloud is divided into bounding boxes.For the grid where feature points are located,all points in it are reserved;for the grid without feature points,the closest point to the center of gravity of the grid is reserved and the rest points are deleted.In the process of simplification,the algorithm can retain the sharp feature points in the point cloud.For the point cloud with fewer sampling points,the surface quality will be poor because the surface information is less,so an upsampling method is studied.Firstly,for feature points,the angle between the normal vector of the point and the normal vector of the neighboring point is calculated.The mean coordinates of several points whose angle is less than the threshold value are added to the point set.This algorithm can make the surface reconstructed by Poisson method fit the surface of the model more effectively and restore the morphological characteristics of the original model more effectively.In order to improve the reconstruction accuracy of the plane projection method on sharp surfaces,it is combined with the region-growing method.After the point cloud carries on the plane projection method to get the triangular plane set,it needs to process the triangular plane.First calculate the normal vector of each point with the method of neighborhood points of vector angle,if less than the threshold,the triangles in the retention,otherwise,the region growing method is adopted for the filtered to choose triangles,use face angle of maximum principle in the process of screening to determine optimized,finally,the points was not involved in the growth region-growing method is adopted to grow.Compared with Poisson algorithm and region-growing algorithm,this algorithm has a good effect on the reconstruction of sharp surface,and can restore the details of the model well.In order to improve the reconstruction accuracy of the Poisson method on sharp surfaces,the Poisson method is combined with the region-growing method.After the point cloud is reconstructed by Poisson method,the triangles whose vertices are in the region of feature points are deleted,the growth edges of the remaining triangles are extracted,and the region-growing method is used to grow them.This algorithm is better than Poisson method in the reconstruction of sharp surface.Using VC++ language to write point cloud display system,using MFC framework and OpenGL graphics library to achieve visual system interface and point cloud and grid rendering.