Lane Detection And Departure Warning Based On Vision

In the driving process, due to the lack of concentration or dozing driver fatigue and otherreasons, it is likely to cause deviation from the original vehicle travel lane.Lane departurewarning system is a kind of method to assist the driver through the alarm to avoid trafficaccident because of lane departure.System can provide lane departure warning intelligently, toprovide the driver with a sufficient reaction time, greatly reducing automobile accidents arecaused because the lane departure.Firstly,Analysis of the research status of the lane departure warning and understand thecurrent research on the system.In order to find out the components,we elaborated the conceptof machine vision systems.Then,Conducted a needs analysis and module division for thissystem and detailed design of the foundation for the subsequent system.In the software systemdevelopment,including pre-acquired image of the road(gray-scale,image smoothing,edgedetection,image segmentation threshold and so on).Removing the interference information toachieve improved system accuracy and timeliness of results.Detailed analysis of differentmethods to achieve the detection of lane,include Hough based lane detection algorithm,basedon least squares method to detect lane,lane detection improved median intercept method.Thencompared the effect of various methods of real-time and selected the better real-time ofimproved median intercept method to completed this article lane detection and fitting. Theequation of the fitted line of the lane and in combination with appropriate parameter selectiondetermines the lateral position of the vehicle based on the lane and the current driving lanedeparture warning parameter decision algorithm.This system using the above technique, combined with roads and image of the softwarehas been tested, the test results show that the software is capable of detecting images ofvarious road and weather conditions, to achieve the desired functionality.