An adaptive approach to lane markings detection

A novel algorithm is represented to determine the position of markings in the presence of distracting shadows, highlight, pavement cracks, etc. RGB color space is transformed intoI₁I₂I₃ color space and I₂ component is used to form a new image with less affection of the clutter. Using an improved edge detection operator, an edge strength map is produced and binarilized by adaptive thresholds. The binary image is labeled and circularity of all connected components is calculated. The self-organizing maps is adopted to extract regions that imply potential markings. Finally the position of markings is obtained by curve fitting. Here color information is utilized fully and all thresholds are set adaptively. The method based on circularity of connected component shows its outstanding robustness to lane markings detection and has a wide variety of applications in the areas of vehicle autonomous navigation and driver assistance system.