Background Filtering and Vehicle Detection with Roadside Lidar Based on Point Association

The high-resolution traffic data (HRTD) is important to intelligent transportation systems (ITS). The roadside LiDAR (light detection and ranging) sensors can provide HRTD by collecting real-time 3D point clouds of surrounding objects. To analyze HRTD, background filtering and vehicle detection are the essential steps. This paper presents an algorithm to extract background and detect vehicles based on point association in 3D point clouds. To improve the accuracy, DBSCAN is combined with time-limited window for vehicle detection according to the relationship between vehicle points.