Video-Based Lane Detection and Tracking during Night

Lane detection plays a vital role in lane departure warning, lane change assistance, and collision avoidance systems. Due to the complexity of night imaging, the lane detection rate at night is usually lower than during the day. In this paper, a new algorithm is proposed for video-based lane detection and tracking at nighttime. The video-based system deals well with images of low gray and illumination variations. First, the difficulty of lane detection during the night is analyzed, and the Gabor filter operator is utilized for image preprocessing. Second, an adaptive splay region of interest (ROI) is established based on the geometry of lane markers. Third, the modified Hough transform is used to detect lane markers. Finally, the Kalman filter is applied for the lane tracking. Experiments based on MATLAB show that the average detection rate of the proposed algorithm is 97.3% and the processing time per frame is 0.22 s.