基于边界的车道标识线识别和跟踪方法研究  

Research on Linear Lane Mark Identification and Track Method Based on Edge

作　　者：王荣本[1] 余天洪[1] 顾柏园[1] 郭烈[1] 

机构地区：[1]吉林大学交通学院,长春130025

出　　处：《计算机工程》2006年第18期195-196,239,共3页Computer Engineering

基　　金：国家自然科学基金资助项目(50175046);中国博士后科学基金资助项目(2004036397)

摘　　要：在道路图像中,为了得到较理想的车道的标识边缘,该文采用基于LOG算子边缘增强的方法得到梯度图像,利用SUSAN算子对该梯度图像进行分割,实现了车道标识线边界的提取。在分割后的梯度图像中,利用Hough变换提取出车道标识线参数,完成了对道路图像中的车道标识线的识别,采用建立梯形的感兴趣区域的办法,实现了对车道标识线的实时跟踪。试验结果表明该方法具有较强的鲁棒性和较好的实时性。In order to get an ideal lane marks＇ edge in a road image, an enhance edge method based on LOG operator is used to acquire a grads mage, then the grads image is divided up by SUSAN operator, the lane mark edges is picked-up, in the divided up image, it is realized to identify lane marks because lane mark line parameters can be acquired by trapezoid areas. Experimental results show that the proposed method Hough transform, the lane mark real-time track is performed by establishing is characterized by reliability, good robustness and real-timeness.

关 键 词：交通运输系统工程 LOG算子 SUSAN算子 车道标识识别 车道标识跟踪 

分 类 号：U495[交通运输工程—交通运输规划与管理]