Volume XLII-2/W7
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLII-2/W7, 825-830, 2017
https://doi.org/10.5194/isprs-archives-XLII-2-W7-825-2017
© Author(s) 2017. This work is distributed under
the Creative Commons Attribution 4.0 License.
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLII-2/W7, 825-830, 2017
https://doi.org/10.5194/isprs-archives-XLII-2-W7-825-2017
© Author(s) 2017. This work is distributed under
the Creative Commons Attribution 4.0 License.

  13 Sep 2017

13 Sep 2017

AUTOMATIC EXTRACTION OF ROAD MARKINGS FROM MOBILE LASER SCANNING DATA

H. Ma1,2, Z. Pei3, Z. Wei1,2, and R. Zhong1 H. Ma et al.
  • 1Beijing Advanced Innovation Center for Imaging Technology, Capital Normal University, Beijing 100048, China
  • 2Chinese Academy of Surveying and Mapping, Beijing 100830, China
  • 3Beijing Eastdawn Information Technology Co. Ltd., Beijing 100190, China

Keywords: MLS data, Extraction, Road Markings, Intensity, Region Growing, Template Matching

Abstract. Road markings as critical feature in high-defination maps, which are Advanced Driver Assistance System (ADAS) and self-driving technology required, have important functions in providing guidance and information to moving cars. Mobile laser scanning (MLS) system is an effective way to obtain the 3D information of the road surface, including road markings, at highway speeds and at less than traditional survey costs. This paper presents a novel method to automatically extract road markings from MLS point clouds. Ground points are first filtered from raw input point clouds using neighborhood elevation consistency method. The basic assumption of the method is that the road surface is smooth. Points with small elevation-difference between neighborhood are considered to be ground points. Then ground points are partitioned into a set of profiles according to trajectory data. The intensity histogram of points in each profile is generated to find intensity jumps in certain threshold which inversely to laser distance. The separated points are used as seed points to region grow based on intensity so as to obtain road mark of integrity. We use the point cloud template-matching method to refine the road marking candidates via removing the noise clusters with low correlation coefficient. During experiment with a MLS point set of about 2 kilometres in a city center, our method provides a promising solution to the road markings extraction from MLS data.