Rapid extraction and updating of road network from airborne LiDAR data

This paper describes an unsupervised approach for efficient extraction of grid-structured urban roads from airborne LIDAR data. Technically, the approach consists of three major components: 1) terrain separation from DSM and classification of ground features, 2) road centerline extraction from generated road candidates images, and 3) completion and verification of complete road networks. A ground-height mask is produced by removing elevated objects from depth image. Then from the mask-superimposed intensity image, road features are segmented out by EM algorithm. This is followed by road centerline extraction from the segmentation image using total least square line fitting approach, during which we develop a Radius-Rotating method to detect road intersections. After that, missing roads inference is executed on road centerline vector map according to gestalt laws. To facilitate inference process, a direction-based cumulative voting technique is developed to evaluate reliability of each road segment. Finally, inferred road features are back projected onto depth and intensity image to test their validity.