Vehicle-based LIDAR for Lane Line Detection

Abstract

Abstract A lane line detection method based on adaptive thresholding of raster map is proposed for assisting driving by fitting lane lines in the process of automatic driving in ports. Firstly, the original laser point cloud is divided into regions of interest to retain the point clouds containing lane lines as well as to remove irrelevant point clouds. Secondly, the raster map is divided based on the intensity of the point cloud, and the ground points are filtered as well as the point cloud of the lane line part is retained. Then the lane lines are segmented by adaptive thresholding for the retained grid, and finally RANSAC is used to fit the lane lines. The experiments prove that the method can extract the lane line features well.