Hongru Hou / Changsha University of Science & Technology;School of Compter and Communications Engineering
Pute Guo / Changsha University of Science & Technology;School of Compter and Communications Engineering
Bin Zheng / Changsha University of Science & Technology;School of Computer and Communication Engineering
Junjie Wang / Changsha University of Science & Technology;School of Compter and Communications Engineering
It is difficult to extract the lane lines accurately in the current auxiliary driving system due to the complexity of the driving environment. In this paper, a new detection method which provides an improved accuracy is proposed. Firstly, a deep learning network of the Unet is adopted to get the potential lane lines. Secondly, the Canny edge detection and Hough transform are used to fit the vanishing point. Thirdly, the position of the vanishing point is used to segment the region of interest (ROI). Finally, the slope of the lines and the relationship between front and back frames in the video are used to select the lane lines. The experimental results show the effectiveness of the proposed method.