Extracting road edges from MLS point clouds via a local planar fitting algorithm

As the basic element of a road, road edges are of great significance for intelligent transportation and urban foundational geographic information construction. Mobile laser scanning (MLS) provides an effective way to extract road information, but it is difficult to extract accurate road edges from a large-scale dataset with complex road conditions. In this paper, we propose a method to extract road edges from MLS data based on a local planar fitting algorithm. First, scanning lines are extracted based on the horizontal projection distance between the laser points. Second, a planar fitting method is adopted to extract road curb points. Road curb points are then clustered and optimized by differentiating the distance between road curb points and the auxiliary line. Finally, a linear least squares fitting method is applied to obtain the road edges. Three experimental datasets with multi-type road markings were used to evaluate the performance of the proposed method. The results demonstrate the feasibility and effectiveness of the proposed method.