A Fast Road Image Segmentation Algorithm Based on Cellular Neural Networks

The main factors that affect segmentation of unstructured road images are shadows and water marks on the road surface. Taking advantage of the parallel image processing capability of cellular neural networks, a fast algorithm for road image segmentation based on cellular neural networks was proposed. In the algorithm, gray threshold segmentation, dilation and erosion, and edge detection using CNN are performed successively. Experimental results demonstrated that the algorithm has strong environmental adaptability, which can fast segment structured and unstructured roads. The proposed method can segment the lane area quickly, effectively and robustly, and can eliminate the influence of shadows and water marks on the segmentation of road images.