his paper presents a novel approach that combines LIDAR and Stereo-cameras. First, a stereo camera is used to guide the extraction of road surface points from a 3D point cloud obtained by a LIDAR. Then, the number of points is reduced based on their elevation and illumination invariant colors. Finally, a post-processing is applied to have a projection in an image to obtain the final segmentation of the road. The method has been tested on three types of roads as defined in the KITTI benchmark. The obtained experimental results prove the effectiveness of our approach and its robustness in accuracy compared to existing approaches in the literature.