The Research And Implementation Of The Detection Technology Of Vehicle Pressing Line And Changing Lane Based On Onboard Camera

With the gradually improvement of transportation infrastructure and the rapid growth of vehicle ownership year by year, traffic accidents are also increasing. Traffic safety problem is becoming increasingly prominent. The vehicle running state detection system based on on-board camera is an important branch of vehicle safety assistant driving. In recent years, with the rapid development of information technology and computer technology, it has become a research focus in the field of traffic safety.It processes the on-board image, gets the traffic parameters, estimate vehicle’s running state according to the traffic parameters, and sends a warning,or even, assist to drive when the vehicle is in a dangerous state. At present, the difficulty technology problem of vehicle running state detection based on on-board camera is the lane line detection.This paper mainly studies the key technology (lane line detection) in the process of vehicle.Firstly, in order to improve the quality of original onboard image, image preprocessing module is studied. Research is mainly focused on analyseing the characteristics of onboard image, setting up the initial region of interest (ROI), converting color image into gray image and enhancing gray image. Finally, a set of image preprocessing algorithm which is suitable for this subject is settled.Secondly, research is mainly focused on the lane line edge extraction algorithm, the lane line parameters extraction algorithm and the lane line parameters tracking algorithm. By comparing traditional lane line edge extraction algorithm and steerable filter edge detection algorithm, the steerable filter edge detection algorithm which can strengthen the lane line edge information is chosen for edge detection. In order to prevent the lane line parameters from mutating caused by detection error, kalman filter is used for tracking and correcting lane line parameter.Thirdly, the vehicle’s state of vehicle line pressing and lane changed is studied on the basis of the lane line extraction. Through the analysis of vehicle line pressing model and vehicle lane changed model, a method of vehicle line pressing which is based on lane line angle and a method of vehicle lane changed which is based on vehicle lane has been put forward.Finally, a system of vehicle line pressing and vehicle lane changed detection based on onboard camera is developed and tested to verify the validity of the algorithm. Test results show that the algorithm studied in this paper can cope with various different scenarios, the lane line detection has high real-time performance and strong robustness, vehicle line pressing and lane changed detection has high accuracy.