基于车载LiDAR点云的城市道路边界提取OA

The vehicle-mounted LiDAR plays a pivotal role in the field of urban 3D modeling due to its efficient 3D data acquisition capabilities.However,in the process of urban road boundary extraction based on vehicle-mounted LiDAR point clouds,obstacles along the roadside,such as vehicles and pedestrians,often pose challenges,leading to the emergence of false boundary points and discontinuities in boundary lines.To tackle this issue,we introduced the elevation standard deviation as a constraint and proposed a road boundary extraction method based on vehi-cle-mounted LiDAR point clouds.Firstly,we designed a data preprocessing scheme,encompassing point cloud segmentation,ground point filter-ing,and scanline separation.Secondly,we constructed a continuous window,and in conjunction with the bidirectional moving window method for scanlines and constraints involving elevation differences,angles,and elevation standard deviations,achieved effective screening of candidate curb points.Subsequently,we employed the DBSCAN algorithm for clustering and denoising to ensure the continuity and accuracy of curb boundaries.Lastly,to address breakpoint issues on the boundaries,we conducted a combined analysis of cumulative curvature and distance infor-mation,enabling precise identification of intersection locations,and utilized a quadratic polynomial curve to connect these breakpoints,resulting in the construction of a high-precision mathematical parameter model for road boundaries.Experimental data demonstrate that,even in complex and variable urban occlusion scenarios,the proposed method achieves over 80%boundary extraction accuracy,validating its efficiency and practi-cal value.