Research Of Lane Departure Warning System Based On Machine Vision

Lane Departure Warning System (LDWS) is an important research direction of vehicle driving assistant technology and an integral part of Intelligent Transportation System. With more and more public attention attracted by safety traffic, vehicle safety driving assistant system also becomes the hot and cutting-edge research.This paper studies the Lane Departure Warning System based on machine vision. The main purpose of the system is to assist the driver to keep the vehicle moving in the right lane and remind the driver of the lane deviation in time so as to avoid the lane deviation caused by the driver's fatigue, neglect or other reasons, reduce the traffic accidents caused by lane deviation and improve the active safety of vehicles. The lane tracking technology in this paper is also the key part of Intelligent Vehicle Navigation. Focused on Lane Departure Warning System, this paper studies the image analysis based on machine vision, lane tracking detection and alarm algorithm.The paper introduces the background, research status and significance of the subject. It also introduces a series of algorithms chosen to run the system. In the process of image analysis, simulate image collected under bad situation by adding noise. Contrast and analyze the result, choosing the median filtering. In the edge extracting process, after comparing various algorithms this paper chooses the Sobel operator for its small amount of computation, effective edge processing and best effect even under the noise.In the lane detection, this paper chooses the Hough transform with good robustness, strong noise immunity and the capacity of checking several lines at the same time. In order to overcome the large amount of computation and bad real-time character of Hough transform, this paper make use of the prediction feedback of Kalman filter to reduce the computation amount of Hough transform. Alarm algorithm combines the advantage of vehicle location and TLC method, so that alarm information more effectively.While choosing the algorithm, this paper runs the simulation using MATLAB. Run the system simulation using the MFC framework and C++ in VS2005 development tools. This paper collects the video when the vehicle is moving, reads the image data per frame and verifies that the selected series of algorithm are feasible and effective by simulation.