Abstract: Extracting accurate road information is of vital importance in urban planning and digital city mapping. We used high-density point clouds collected by mobile laser scanning(MLS) systems, and proposed a novel approach based on road boundary detection in curb missing areas. First, the road was roughly extracted by using plane detection algorithm. Next, we analyzed the difference in laser reflection intensity and geometry attribution between road boundary and two sides of surface and retrieved the point clouds of boundary by setting reasonable threshold. Finally, the road surface was accurately extracted based on curve fitting of discrete points. To evaluate the performance of the proposed method, experiments were conducted using urban point clouds data. The extraction method achieves correctness, completeness, and quality rates of above 90%, which indicates that the proposed method effectively detects road boundary and extracts road surface accurately in curb missing urban areas.

Key words: mobile laser scanning, road extraction, point clouds, reflected intensity