Classification of Point Cloud Using Received Light Intensity According to The Degree of Separation

In the conventional intersection recognition method, shape information obtained from 3D LiDAR is used. In order to recognize intersections composed of the road and the others, it is necessary to classify obtained point cloud. However, there is almost no difference in shape between grass and asphalt. For this reason, it is difficult to distinguish them by using the shape information. Therefore, we use received light intensity to broadly classified into grass and asphalt. In this paper, we detect points on the grass by comparing the received light intensity at the equidistant points from the 3D LiDAR. The appropriate threshold for distinguishing materials is automatically calculated for each distance using Discriminant Analysis Method according to validity of classification. Our proposed method will provide that extracting of points on the grass zone is robust and appropriate for the changing reflection intensity of the ground surface.