Research On Building Facade Extraction And Modeling Based On Ground LiDAR Data

Buildings as an important goal of urban 3D modeling,its location border information in the current map updates and navigation,real estate planning and other applications will play an important role.Terrestrial 3D laser scanner can obtain a large area of high-resolution surface of the measured object mass of three-dimensional data,but the original 3D laser scanning data in addition to the building facade point cloud,also contains other non-building point cloud data such as the ground,trees,pedestrians,vehicles and the city part of the infrastructure point cloud and other noise points,need to extract the building facade information,in order to carry out 3D modeling of the target building.Therefore,how to extract the building facade quickly and accurately from a large number of point cloud data,has become a hot topic for many scholars to study,is an important part of work in the "digital city" building,The accuracy of the reconstruction determines the reconstruction accuracy of the late model of building.Random Sample Consensus as a robust algorithm,it is still able to get the desired processing results in the data error rate of more than 50%,it can effectively suppress the impact of noise,extract the correct feature lines and feature surfaces,is a robust and efficient method of fitting the mathematical elements from the sample,It is widely used in the field of computer vision,such as basic matrix estimation,feature matching and motion model selection.RANSAC algorithm is currently used for airborne and car LiDAR data processing,there are few studies in the data acquisition of terrestrial 3D laser scanners.The traditional RANSAC algorithm needs to determine the threshold in advance,In the process of building facade information on the plane point cloud extraction process,the selection of thresholds has a great effect on the accuracy of plane extraction.Based on the brief introduction of the terrestrial 3D laser scanning technology,expounded the principle of RANSAC algorithm,aiming at the characteristics and shortcomings of the traditional RANSAC algorithm,improved the traditional RANSAC method based on building facades extraction,The RANSAC algorithm is optimized based on point cloud density and radius density.Through the experiment to complete a building facade extraction.Two different modeling methods using surface reconstruction method and parameter method are studied.Respectively,to obtain the two point cloud data model reconstruction.The concrete steps of the two model reconstruction methods are described.and the two modeling techniques were used in combination.Finally,two modeling techniques are compared and analyzed.