Research On Point Cloud Processing And Reconstruction Of Complex Surface Measurement Data

With the development of modern design and aerodynamics,the geometric structure of machining parts becomes personalized and complex.Due to the continuous development of optical measurement,the accuracy and efficiency of laser 3D scanning technology are greatly improved.In this paper,a typical turbine blade is taken as the object,and the principle and effect of various de-noising algorithms are obtained through the experimental analysis of the point clouds collected by the self built three-dimensional laser scanning system.The sorting algorithm and fast splicing and merging algorithm for scattered point clouds are proposed,and the surface reconstruction algorithm and hole repair algorithm for scattered point clouds are studied,At the same time,the software instantiation verification analysis is written.The significant research contents and main conclusions are as follows.In depth explore the blade measurement data of a variety of noise removal algorithm,using a combination of denoising algorithm to remove the original point cloud in a variety of miscellaneous points.According to the characteristics of disordered data points in scattered point cloud,this paper studies the distribution characteristics of grid and linear point cloud,and puts forward inter layer sorting and bead chain sorting algorithm,which provides data basis for subsequent point cloud fast retrieval and boundary point processing.In order to get the complete point cloud data,this paper explores the point cloud splicing algorithm based on ICP,and discusses the global ICP splicing algorithm of coarse and fine registration and the splicing algorithm of local participation in calculating the output splicing matrix.By calculating the two similar point clouds obtained from the small angle change of the measured object,the rotation and translation matrix of the measured data is obtained and saved,and then the multi angle stitching algorithm is realized with this angle as the benchmark and multiple rotation acquisition.A redundant point removal algorithm based on nearest neighbor search is proposed to remove a large number of redundant and overlapping points in the overlapping area.Based on the plane projection method and implicit function method,the surface mesh of scattered point cloud is reconstructed,and the smooth and complete 3D model of blade profile is obtained.This paper analyzes the hole repair algorithm based on radial basis function,completes the closed hole detection and boundary line division,and carries on the experimental verification to the hole boundary detection and repair in the smooth surface.In summary,this paper uses vs2013 and qt5.11.2 to build the development environment,uses C++ language to design and develop the point cloud processing and surface reconstruction system software,and build the system UI by QT.Taking the steam turbine blade point cloud as an example,the feasibility and practicability of the proposed algorithms are displayed and verified visually.