Research On Lane Detecting And Lane Departure Warning Technology Based On Machine Vision

With the rapid growth of vehicle ownership in China,road traffic safety has become an issue of increasing concern.Most of the road traffic accidents are caused by the driver's negligence,many of which are caused by the lateral displacement of the vehicle.The lane departure warning technology can issue an alarm when the vehicle is laterally deviated to remind the driver to correct the driving behavior in time,thereby effectively avoiding traffic accidents caused by lateral deviation of the vehicle.The research on the lane departure warning system mainly focuses on two key technologies of lane line detection fitting and deviation warning strategy.The main research work of this paper is as follows:(1)The road image preprocessing module proposes two-step ROI extraction method,grayscale method based on color component and improved Otsu binarization method,and carries out ROI extraction,grayscale,image smoothing,edge extraction,binarization and other preprocessing operations.(2)Based on the shortcomings of Hough transform,a constrained probabilistic Hough transform detection algorithm is proposed to detect linear lane lines,and the improved RANSAC algorithm and parabolic model are used to detect curved lane lines.In order to more accurately adapt to the actual driving environment,a linear-parabolic model based on dynamic division of the near and far field of view is proposed.(3)A lane departure warning model based on the current lateral offset of the vehicle and the slope of the lane line is proposed,and the model is verified by real vehicle experiments.The Jingyi S50 EV vehicle experiment and the loading and recording road video verification show that the core algorithm can detect the lane line very well and can issue early warning when the vehicle deviates,which has better robustness and real-time performance.