基于线性抛物线模型的车道检测与跟踪方法  

Lane Detection and Track Based on Linear-parabolic Model

作　　者：洪敏[1] 周鸣争[1] 梁祥君[1] 修宇[1] 

机构地区：[1]安徽工程大学计算机与信息学院,安徽芜湖241000

出　　处：《长江大学学报（自科版）（上旬）》2013年第3期53-57,6,共5页JOURNAL OF YANGTZE UNIVERSITY (NATURAL SCIENCE EDITION) SCI ＆ ENG

基　　金：安徽省教育厅自然科学基金重点项目(KJ2007A046)

摘　　要：车道检测是车道偏离预警系统的前提,是车辆辅助驾驶系统的重要组成部分。提出一种基于线性抛物线模型车道检测与跟踪方法:首先,提出一种新的车道线检测感兴趣区域(ROI)设置方法;其次,引入双阈值算法对道路边缘进行提取,利用Hough变换初步获得车道中线及消失点坐标并提取道路标识线;最后,建立线性抛物线模型,引入卡尔曼滤波(KF)对车道线进行跟踪,并用卡尔曼滤波对模型参数进行更新。多段实地采集的视频仿真试验结果表明,该算法能快速、稳定的检测和跟踪车道线。Lane detection,which is the premise of lane departure system,plays an important roll in vehicle driver assistance systems.This paper proposes a lane detection and track based on linear-parabolic model.Firstly,it is introduced puts forward a new region of interest(ROI) settings.Secondly,a dual-threshold algorithm for road edge extraction,the Hough transform preliminary obtains lane center line and vanishing point coordinates and extracting the road marking line are obtained.Finally,the linear-parabolic model is established by using the Kalman filter(KF) to track the lane line and update the model parameters by using the Kalman filter.Through the multiple field acquisition of video,the simulation experiment results show that the algorithm can detect and track lane line quickly and stably.

关 键 词：车道检测 线性抛物线模型 感兴趣区域 双阈值 卡尔曼滤波 车道线 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]