Road markings recognition using image processing

Repainting road markings is an important road maintenance task, An automatic repainting module (called << MRA >>) is being developed in France by the Laboratoire Central des Ponts-et-Chaussees. In this context, our goal is to design image processing methods for automatically extracting the parameters of the repainting process (e.g. line width or dashes modulation) from the scene images. We detail in this paper a three-level road markings recognition algorithm. Firstly, a road sides extraction procedure reduces the analysis area to the roadway and a road markings detection is performed in this area. Secondly, the segmentation is simplified using a projection-based method, leading to a polygonal representation of road markings. Lastly, the parameters to be transmitted to the repainting system are extracted.