Target Recognition Tracking Based on 3D Laser Point Cloud and Traffic Safety Evaluation of Expressway Merging Area

Based on roadside LiDAR data, a roadside-based target recognition and tracking method is designed for the problem of extracting traffic target trajectories, which accomplishes streamlining and denoising, ground segmentation, target clustering, and matching tracking of point cloud data. It also proposes voxel raster filtering and ground plane fitting algorithms to reduce the computation amount, correct and update the preliminary prediction value by Kalman filtering, and obtain the final target tracking results. Finally, the data collected from UAVs in a provincial capital city are adopted for example verification, which proves that the proposed method can effectively accomplish the preset goals, and is scientific, advanced and robust.