Lane Detection and Detection-Error Correction by using Hough Transform

Abstract

Intelligent vehicles with robust lane detection and warning systems (LDWS) have been gaining popularity for enhanced safety. Robust lane following is important to warn the driver efficiently from lane departure. Accurate lane detection under diverse road and weather conditions is a critical issue for LDWS. In this paper, we have developed a cost effective vision based approach which is capable of detecting the lanes with robustness and capable of overcoming the errors caused due to the illumination changes or clutters. A cost effective technique using a single camera has been used to capture the road. The edges are detected by Sobel edge detector. A filtering process is used to remove noises e.g., shadows and clutters in the edge map. The Hough transform is used to find the lanes in the filtered edge map. The online tracking of lanes is done by using data comparison through Hough transform.