Vectorizing urban road markings from mobile laser scanning point clouds

Abstract

Elaboration of topographic maps from Mobile Laser Scanner (MLS) point clouds is a long and tedious manual task. This study presents a method that automates the underlying vectorization process. We propose a road marking vectorization method that can operate on road urban areas acquired from a MLS. The presented technique is based on a bottom-up contour-based technique incorporating prior knowledge at different stages.Our experiment shows that the use of "road logic" is a key element in the success of the detection and vectorization process.