Research On Point Cloud Filtering And Plane Pitting Algorithm Based On 3D Laser Scanning Technology

With the rapid development and widely application of 3D laser scanning technology,Point cloud data processing based on 3D laser scanning technology has become a hot and difficult research.Point cloud data processing is generally included Point Cloud Stitching,Point Cloud Compression,Point Cloud Segmentation,Point Cloud Filtering and Point Cloud Plane Fit,in which Point Cloud Filtering and Point Cloud Plane Fit are the keys to the study.The study of this paper is based on the point cloud data collected by Leica ScanStation C10 3D laser scanner.The research contents include point cloud data acquisition and all aspects of Point Cloud Data Preprocessing,which the mainly studies is algorithm of Point Cloud Filtering and Plane Fitting algorithm.The following is the main content of this paper:(1)This paper focuses on the Iterative Least Squares Filtering Algorithm,for the reason that it can not effectively eliminate low noise and its Weight Determination Function is unreasonable,In this regard,this article has done some improvements.First,we use the method based on point neighborhood to eliminate low noise points,then we grid block the point cloud data unevenly,after that setting different offset values according to the noise situation in the block,In the next step,fitting the trend surface based on improved full value,Finally,separating noise points and pavement points.Experiments show that the new algorithm has better filtering effect by contrasting with conventional mean filtering,median filtering,least squares filtering,Laplacian algorithms,and the original algorithm.(2)This paper profoundly analyzes TLS and RANSAC——the two of the most commonly used and well-performing planar fitting algorithms.Taking into account their advantages and disadvantages,this paper presents a new fitting method.First,using RANSAC to fit the point cloud,finding the best parameter estimation model through multiple iterations,this step can effectively eliminate outliers in point cloud data.Then,using TLS to fit the model,finally,the fitting plane closest to the ideal model is obtained.Experiments show that the new algorithm has better Point cloud fitting effect by comparing with the original single algorithm.