Research On Building Information Extraction Based On Airborne LIDAR Data

As the inevitable trend of knowledge economy and information society,"digital city" represents the new world trend and the direction of urban development.As one of the most important elements of urban area,building information extraction and analysis is becoming more and more important.In recent years,many scholars in the field of photogrammetry and remote sensing have launched the research of building and building information based on airborne LiDAR point cloud data(half)automatic information extraction.Airborne LiDAR is a new type of active telemetry instrument,which can quickly and directly obtain mass,high precision and high density surface 3D data.It is a 3D spatial data acquisition technology with efficiency,precision and economic advantage.Therefore,airborne LiDAR data gradually becomes the data source which is difficult to replace in building information extraction and model reconstruction.This paper takes the Linyi city of Shandong province as the research area,and uses airborne LiDAR aerial data to study and test the problems of noise removal,filtering classification,building point cloud extraction,building contour extraction and contour line regularization,which are related to building information extraction,and the specific contents are as follows:(1)The airborne Li DAR point cloud is organized by the three-dimensional grid data structure,and the discrete and cluster noise points outside the main body of the point cloud are removed according to the neighborhood relationship between the grid and the region growth algorithm.(2)According to the rule grid structure,the lowest point in the grid is selected.In order to build the TIN,the point cloud distance from the ground is obtained for the seed point.Then the LiDAR point cloud is divided into the ground point and the non ground point through the balance filtering of height deviation,and the non threshold filtering is realized.(3)Using the multiple echo information of the non ground point cloud and the height difference to eliminate the non building points,the fitting degree of the point cloud within the grid range is detected by the least square plane fitting,and the buildings that meet the specified threshold area are screened through the detection of the Unicom domain.Aiming at the misjudgement situation of building point cloud grid edge,the height difference and height difference of data points in neighborhood grids are screened to improve classification accuracy.(4)Combined with the virtual grid,the data acquisition scope of traditional Alpha-Shape algorithm is improved,and the contour of building point cloud is generated quickly.The key points of buildings are extracted by the method of angle change monitoring,and the accuracy of regular contour is improved by intersecting the contour points between key points.The point cloud classification results obtained by the above algorithm are compared with the processing results of commercial LiDAR data processing software Terrasolid and LiDAR360.The overall error of the classification is similar,which verifies the effectiveness of the algorithm.