An approach to detect and delineate street curbs from MLS 3D point cloud data

• We propose a method to detect street curbs from 3D point clouds.
• Image processing techniques are used to detect linear features.
• The method is valid for both straight and curved road sections.
• Occluded areas between straight road sections are detected.
• The method determines the upper and lower edges of detected curbs.

This paper is focused on the detection and delineation of curbs and street boundaries using a 3D point cloud registered by a mobile laser scanner (MLS) system. Non-manual roadside detection is an important issue in road maintenance, 3D urban modeling, and autonomous navigation fields. The proposed method is able to detect upper and lower curb edges and delineate even occluded or hidden roadsides in most environments. It is based on a rasterization and segmentation approach for curb detection while preserving the reference to the original point cloud avoiding loss of information.The method was tested using two point clouds from various street scenes measured by an Optech Lynx Mobile Mapper System, providing accuracies in curb extraction over 95% in both datasets.