Publication

Abstract : One of the major challenges on highways is to avoid an unintended departure from the lane. This paper proposes a lane departure warning system with the help of monocular vision. The efficiency of such a system is subject to clarity of lanes, weather conditions and also method of acquisition. This paper proposes a method of lane detection that is robust to stray edges within the frame. Canny edge detection is utilized on the pre-processed images to obtain maximal intensity edges, followed by lane detection using Hough transform. In this paper we try to utilize selective property of the edges obtained from canny edge detection to reduce noisy edges and improve the false positive rate. The system has been tested to be effective in fully illuminated as well as badly illuminated road conditions with satisfactory results. The average detection rate obtained is 95%.