A DYNAMIC VISION ALGORITHM TO RECOVER THE 3D SHAPE OF A NON-STRUCTURED ROAD

This paper describes a method in which an on-board monocular monochromatic camera is used for locating a vehicle on a non-marked road and recovering its 3D geometry. The method first uses an image segmentation stage to deduce the road area in the image. In the next stage, the vehicle location is estimated by assuming that the road is straight near the vehicle. In the last stage, the 3D geometry of the road is recovered. Results are presented for each stage on a non- structured road and a forested road.