Automated Extraction of Road Markings from Mobile Lidar Point Clouds

Abstract

This paper proposes a method for extracting road markings from mobile lidar point clouds. Among implicit information of laser points, the strength of reflection of laser points is closely related to the property of road materials that helps in detecting road markings. An interpolation method is used first in order to generate a georeferenced feature image of the point clouds, which helps to isolate the points of road surfaces. An algorithm is used to separate these points within a range according to their strength of reflection. The separated points are further segmented to remove non-road points based on their height threshold. The outlines of road markings are extracted from the segmented points using the semantic knowledge of road markings. The results presented in the paper demonstrate that the method used was very promising for automatic extraction of road markings from lidar point clouds collected by a land-based mobile lidar system.