An Illumination-Robust Approach for Feature-Based Road Detection

Road detection algorithms constitute a basis for intelligent vehicle systems which are designed to improve safety and efficiency for human drivers. In this paper, a novel road detection approach intended for tackling illumination-related effects is proposed. First, a grayscale image of modified saturation is derived from the input color image during preprocessing, effectively diminishing cast shadows. Second, the road boundary lines are detected, which provides an adaptive region of interest for the following lane-marking detection. Finally, an improved feature-based method is employed to identify lane-markings from the shadows. The experimental results show that the proposed approach is robust against illumination-related effects.