Research On The Automatic Extraction Method Of Laser Point Cloud Building Based On Improved RANSAC Algorithm

Airborne lidar is a three-dimensional direct detection technology integrating laser ranging,global positioning system(GNSS)and inertial navigation.It is one of the most potential geospatial information acquisition technologies in today’s society.It can actively and quickly acquire the dense three-dimensional point cloud data,reflectivity,texture and other information on the surface,making up for the traditional measurement,aerial photogrammetry and other technologies.The shortcomings of means.In the face of massive point cloud data,how to effectively extract buildings,remove vegetation,street lights,roads and other point cloud data unrelated to buildings,and provide basic data for urban mapping,smart city construction and other fields is the most concerned and the most urgent technical problem to be solved.Based on the task of fast extraction of point cloud on the top of building,this paper studies the technology of fast extraction of point cloud based on improved RANSAC algorithm.The main research work is as follows:This paper summarizes the working principle of airborne lidar system and the theoretical method of rapid extraction of building point cloud.This paper mainly analyzes the structure of Airborne lidar point cloud acquisition system and the characteristics of lidar point cloud data,and analyzes the basic workflow of lidar point cloud data extraction from the perspective of integration of internal and external business;expounds the basic principle and implementation process of common building extraction algorithms such as morphological filtering,slope based filtering,regional growth method and random sampling consistency algorithm in detail,and analyzes and discusses four kinds of algorithms The advantages and disadvantages of this method in building point cloud extraction,and determine the building point cloud extraction algorithm RANSAC algorithm to be used in this paper.This paper proposes an improved RANSAC algorithm for building point cloud extraction,and tests it.The traditional RANSAC algorithm is improved from four aspects:least square fitting optimization,data traversal optimization,broken surface optimization and pseudo plane optimization.Based on Visual Studio 2010 integrated development environment,the improved algorithm is programmed and implemented using C + + language and PCL point cloud open source library.The experimental results show that the improved RANSAC algorithm can effectively and quickly extract the shape of buildings,effectively identify the point cloud of building facade and vegetation around buildings,extract the point cloud of two buildings with different shapes in Guangzhou area about 60%,and extract the point cloud of three buildings with different densities in Zhoushan area about 70%.The comparative analysis shows that the building Point cloud extraction is affected by the density of point cloud,surrounding vegetation,noise and other factors.The algorithm designed in this paper can filter out the low vegetation and other noise points around the building,obtain the point cloud information of the building shape stably,and provide technical support for the further application of the point cloud data.