基于激光点云的道路三维数字地图构建方法OA

To address the technical pain points of insufficient extraction accuracy of road marking lines and poor classification performance of non-ground objects in the construction of road 3D digital maps supported by vehicle-mounted LiDAR point clouds,this paper proposes a hierarchical and refined construction method.First,the point cloud data is preprocessed by filtering,and the regional growth algorithm is integrated with the intensity information features of LiDAR point clouds to achieve accurate extraction of road marking lines.Second,the SPVNAS semantic segmentation network is adopted to complete the recognition and segmentation of small targets,and multi-stage clustering combined with model fitting morphology is applied to realize the accurate recognition and segmentation of complex tree objects.Finally,the individual modeling of ground objects is carried out and integrated into the Unity3D platform to construct a stereoscopic and high-precision 3D digital map of urban roads.This method effectively improves the accuracy and efficiency of road marking line extraction,realizes the intelligent classification and individual modeling of non-ground objects,and provides a feasible technical approach for the construction of road digital maps driven by vehicle-mounted 3D LiDAR point clouds.