基于SIFT和SSDA特征匹配的实时车道线检测  

Real-time lane detection based on the feature matching of SIFT and SSDA

作　　者：吕亚运 郎朗[1] 杨会成[1] 

机构地区：[1]安徽工程大学安徽省电气传动与控制重点实验室,安徽芜湖241000

出　　处：《安徽工程大学学报》2015年第4期54-61,共8页Journal of Anhui Polytechnic University

基　　金：安徽省高校自然科学研究重大基金资助项目(KJ2014ZD04)

摘　　要：车道线检测是智能交通系统研究的一个重要方向.提出了一种基于SIFT和SSDA特征匹配的车道线检测算法.首先通过改进Sobel算子来提取车道线边缘特征,并设置感兴趣区域(ROI)缩小范围,然后进一步提取车道线特征点并通过Hough算法拟合成直线,接着采用SIFT算法提取影像关键点并进行初次匹配,最后运用SSDA算法进行影像精匹配,从而融合成完整车道线影像.结果表明,采用改进Sobel算子可以提取到更多的车道线资讯;而采用SSDA算法,比传统NCC算法匹配精度更高,实时性进一步提高.Lane detection is an important research field of intelligent transportation system,and a lane detection algorithm based on the feature matching of SIFT and SSDA is proposed.Firstly,edge features of the lane are extracted by the improved Sobel operator,and the region of interest（ROI）is set to narrow the range.Secondly,the lane feature points are further extracted and a straight line is fitted by Hough algorithm.Thirdly,the SIFT is adopted to extract the key points and do the initial matching.Finally,SSDA is adopted to do further matching and fuse the two images into a complete one.The results show that the improved Sobel operator can extract more lane information.The matching precision is higher than that of the traditional NCC algorithm by SSDA,and the real time is further improved.

关 键 词：车道线检测 SIFT 图像匹配 SSDA 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]