Research On Lane Departure Warning Algorithm Of Intelligent Driver Assistance Systems

With the continuous improvement of people's living standards, the number of people holding cars in China has greatly increased.Due to the increasing number of cars,causing traffic problems have become increasingly prominent,and the traffic accident rate increased significantly.And The traffic accident caused by the driver's visual fatigue or inattention is the most serious.So the Lane Departure Warning System (Lane Departure Warning System, LDWS) is a research hotspot in recent years. Lane departure warning is an important part of the intelligent driver assistance systems.Its main function is when the vehicle will deviate from the lane, lane departure warning system give the signal alarm through sound, light, vibration to alert the driver the deviations from the current vehicle.In this way the system can greatly reduce the incidence of traffic accidents.Firstly, this article introduces the research background and significance of lane departure warning system, analyzes the research status at home and abroad. Then,research on the key algorithm of lane departure warning system.The main work is as follows:(1)Preprocessing of the road image.Because the road in the original image is noisy color image,this causes interference on lane detection and contains large amount of computation, so we need to preprocess the original image.In this paper, the gray image to reduce subsequent processing of the data, then by Gaussian filter, Canny edge detection and two value segmentation algorithm enhance the lane information.Laid the foundation for lane detection and tracking.(2) According to the characteristics of the road image, the thesis firstly uses the combination model of lane.Then designing detection algorithms for different regions::in the near field, the lane is simplified to a linear model, use Hough transform to the lane line detection; in far field, Lane approximate parabola model using least square method to detect the lane.And According to the parameter information of the lane in two regions,we can judge the lane and get the lane.(3) Lane departure decision algorithm mostly needs camera calibration and image reconstruction. Aiming at this question,the article proposes a new deviation decision algorithm.This new algorithm according to the change of lane's vanishing point and lane's slope to determine whether deviating from the vehicle.First of all,according to the slope of lane in the current image to judge whether the vehicle to the left or to the right,and then combined with the change of lane's vanishing point to analyze the kind of deviate, and sends out the alarm signal.This algorithm avoids the complex process of camera calibration, reduces the complexity of the system,and increase the detection rate of the system very well.(4) The article designs and develops software of the system,and introduces the image acquisition and display, the lane line detection, alarm and other functions module realization process in detail.Through the development of software system, the actual application of the algorithm in papers provides experimental platform.