The Theory Of Point Cloud Clustering And Surface Fitting On Virtual Measuring System

Aiming at the deficiency of existing clustering method of point cloud and surface fittingmethod, this assay is proposed to improve clustering method of ant colony optimization projectionpursuit and radial basis function interpolation method based on density. This dissertationconstructs a unified virtual testing system of point-cloud data acquisition, wireless transmissionand storage, and deduces an error propagation model of unified virtual measure point cloudcoordinates based on analysis of error sources and its relations with coordinate systems.The studyputs forward an improved projection pursuit method of ant colony algorithm to realize3D pointcloud cluster, constructing the arctan function which update the coefficient of pheromone in antcolony algorithm. Evaluating optimization results from the angle of convergence of iteration, itevaluates quantitatively the effectiveness of different clustering algorithms, using the method ofDUNN index. It puts forward radial basis function based on point cloud density, discovers thequantitative impact of surface interpolation point and the out-of-plane constraint point towards theradial basis interpolation precision, analyzes the error influence of quadric surface fitting, Nurbssurface interpolation and radial basis function interpolation based on density towards volumecalculation.