Study on curb detection method based on 3D range image by LaserRadar

This paper presents a method to detect curb in city by 3D range information, which is provided by the laser radar mounted in a moving vehicle. First, a CS median filter is used to suppress noises. Then, the 3D information of curb is obtained based on the height difference between curb and road surface. After that, edge operation and mathematical morphology filter are used to segment curb from the whole range image. In this way, the relative position between the curb and the host vehicle can be acquired, which is important to driving-assistance. Experimental results show that the approach has high robustness and accuracy.