基于熵最大化图像分割的直线型车道标识识别与跟踪方法.pdfVIP

23 6 Vol23 No6 2006 6 Journal of Highway and Transportation Research and Developm ent Jun2006 ( 2006) 1 2 1 1 余天洪, 贾 阳, 王荣本, 郭 烈 ( 1 , 130025; 2, 150001) 为了解决在道路路面材料不一致或光照不均匀情况下的车道标识线的识别和跟踪问 , 提出一种基于熵最大化 的图像分割可变形模板的车道标识线识别及跟踪方法该方法结合了图像变窗口处理技术和基于熵最大化分割方法来 实现对道路图像的理想分割, 然后利用可变形模板匹配方法得到车道标识线参数, 最后采用建立梯形感兴趣区域的方 法实现对车道标识线的实时跟踪试验结果表明 该方法具有很好的可靠性鲁棒性和实时性 交通运输系统工程; 一维熵; 图像分割; 车道标识识别; 车道标识跟踪 U4912+ 23 A Linear Lane Identification and Tracking Method Study Based on Maximum Entropy Image Segmentation 1 2 1 1 YU Tianhong , JIA Yang , WANG R ongben , GUO L ie ( 1College of Transportation, Jilin University, Jilin Changchun 130025 , China; 2Harbin Institute of Technology, Heilongj iang Harbin 150001, China) Abstract: To realize lane mark identification and tracking under such condition s as uneven road surface materials and different illumina tion etca lane mark identif ication and tracking method based on maximum entropy image segmentation is proposedThe authors use this method combined w ith image w indow variation technology and maximum entropy segm entation method to divide up the lane m ark of road im age, then to get the lane mark parameters by the m atch technique based on the binorm alized adjustable template, f inally to track the lane mark in realtime by establishing trapezoid areasExperimental results show that the proposed method is ch arac