中文摘要：

为了得到比较理想的道路图像中车道标识的边缘，文中采用了基于二维直方图熵最大化的车道标识边缘提取方法。对此边缘图像采用定向边界跟踪方法提取出车道标识特征点，对这些特征点采用线性回归技术就可以得到车道标识线参数，从而完成对道路图像中的车道标识线的识别，并且文中还采用了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。

英文摘要：

In order to get an ideal lane marks＇ edge in a road image,this paper uses a method which is based on maximum two-dimension entropy to pick-up lane mark edges.Lane mark character points can be found by using fixed orientation edge trace in the edge image.It is realized to identify lane marks because lane mark line parameters can be acquired based on linear regression technique by using these polnts.Finally lane mark lines real-time tracking is realized by applying trapezia AOI method.