Automatic road extraction from mobile laser scanning data

Extraction of road surface and boundary is essential for autonomous vehicle navigation, road monitoring and important scene structures extraction. Mobile laser scanning (MLS) technology as a new information acquiring manner can quickly scan the whole scene and provide density and accurate 3D coordinate data and other information such as trajectory, color and reflectance. In this paper an automatic road extraction method is proposed based on trajectory information from mobile laser scanning data. Through the trajectory, location and approximated direction of local road patch could be determined. Searching algorithm is applied along the approximated road direction and the orthogonal direction. To determine the road boundary, a hypothesis testing method based on local altitude variance is used. To filter false boundary points, local altitude mean value is applied. Experiment results demonstrate the reliability of the proposed algorithm for automatic road surface and boundary extraction.