Building Change Detection Based On Urban Area Of Airborne LiDAR Point Cloud

Buildings are the key elements of urban planning and construction,and the most vulnerable part of the city.Therefore,Detecting changes in urban buildings is of great significance for urban development and planning.The high density and precision of LiDAR(Light Detection And Ranging)point cloud data provide a new data source for building change detection.At present,research on building extraction based on airborne LiDAR point cloud has made good progress,and many effective methods have been proposed.However,There have been few reports on the use of multi-phase airborne LiDAR point clouds for the detection of changes in buildings in urban area.Therefore,this thesis conducts in-depth research on the detection of building changes in urban areas based on multi-phase airborne LiDAR point clouds.The specific research contents and main results are as follows:(1)Building roof surface extraction.The roof surface of the building is separated and the roof surface of the adjacent building is incompletely divided,Edge point segmentation errors and other issues that need to be resolved.For the complete segmentation of the roof surface of the building,Based on the regional growth algorithm,this paper uses the normal vector of the point cloud of the building roof to focus on one direction and the spatial distance from the point to the roof surface.The regional growth algorithm combined with the normal vector is used to extract the roof surface of the building.The experimental results of the two research areas show that compared with the region growing algorithm,the region growing algorithm combined with the normal vector can segment the adjacent and nearly parallel roof surfaces,and the segmentation accuracy of the edge points of similar objects is higher.(2)Building boundary extraction.Using the Alpha-shape algorithm extract the initial contour of the building,then Based on the main direction forced orthogonal algorithm completed the regularization of the building outline,and evaluating the building boundary extraction results.The method is applicable to buildings with adjacent buildings whose boundaries are approximately orthogonal,not all buildings.In view of the fact that the average spacing of the experimental LiDAR point cloud data is 0.25 m,the RMS(Root Mean Square)value of the building boundary is less than 2 times the average point cloud spacing,indicating that the method has higher accuracy.(3)Two-phase point cloud position difference assessment.First determining the ridge line through the adjacent roof surface,then b ased on the invariant feature of the position of the ridge line is used to estimate the spatial position difference of the LiDAR point cloud acquired in different periods.In this study,the rotation angle between the point cloud of 2010 and 2014 is 0.016 degrees,the offset is 0.019m,and the offset is smaller than the experimental LiDAR point cloud spacing,It indicates that the position difference of the two-point cloud data is not much different,and the accuracy after registration is achieved.(4)Building change detection.The change in the plane of the building is through the stack analysis of two regul rized building boundaries to complete.The median value and the mean value of the corresponding roof surface are used to obtain the deformation amount on the height of the building,which will minimize the influence of the difference between the point cloud and the accidental error on the calculation of the deformation amount The results of the detection of changes in the plane and height of the building are analyzed respectively.The building changes in the study area are analyzed,and the number of buildings in the experimental area is statistically analyzed.The algorithm detects a total of 96 building changes,the complete rate reaches 92.1%,and the accuracy rate reaches 93.9%.