Research On The Technology Of Anti-collision Warning System Based On Lane Departure

Advanced Driver Assistant Systems(ADAS)is a kind of auxiliary device for improving the efficiency of driver——vehicle system.In auxiliary driving system,first task is to achieve the driving state,vehicle and environmental parameters.Obtained information is analyzed and processed comprehensively by the vehicle controller.The output signal for the driver can be identified sensory.This can assist the driver to avoid traffic accidents and achieve the purpose of energy saving,safety and environmental protection.In this paper,based on the research of intelligent vehicle system at home and abroad,various algorithms are analyzed and compared,then some specific methods to detect the lane and vehicle information is put forward.The main contents of this paper are as follows:(1)For lane detection,first step is preprocessing the image of the division for region of interest without affecting the detection results.Then the picture is operated by median filter and extracted edge information by Canny operator.Real car line is extracted by the traditional Hough transform with constraints.(2)For vehicle detection,coarse location of target was determined for candidate regions with Otsu algorithm improved by genetic algorithm method.Then Haar feature Adaboost classifier is used to determine whether the target is a vehicle.(3)For vehicle tracking,Kalman filter algorithm and Meanshift algorithm are used to realize the tracking.Through the accurate tracking of the vehicle in front,the position of the target can be determined effectively.So that the target can be located in the case of the sudden disappearance of the target signal to be detected.(4)For the early warning strategy,the TLC method is used to combine the image and information in the lane departure warning.The method of calculating the safe distance based on the vehicle speed is used in the anti-collision warning.Through the geometric relation projection model combined with camera calibration method,the current position of the vehicle,the deviation angle and the distance from the vehicle in front can be calculated.