Abstract: A real-time on-line algorithm to estimation the lane departure time based on Inverse Perspective Mapping（IPM） is proposed in this paper.At first，we obtain the binary lane image by image processing and eliminate the perspective effect using the IPM model，then we detect the lane markings with Hough Transformation and finally compute the time of lane departure to judge whether the car has an underlying lane departure danger.In this paper，we will introduce this method detailedly and give the test results which prove that our method is efficient in detecting the lane departure.

Key words: lane departure, inverse perspective mapping, Hough Transformation, lane markings

摘要： 提出了一种基于图像逆透视变换后的车道偏离时间的实时在线估计算法。该算法先对道路图像进行预处理，得到二值化的道路图像，然后通过逆透视变换方法消除图像的透视效果，用Hough变换方法检测车辆所在车道的左右道路标志线，最后估算车辆偏离车道的时间，判定车辆是否会偏离车道。对该算法进行了详细介绍并给出了实验结果，结果表明该算法能够准确地判断车道偏离。

关键词: 车道偏离, 逆透视变换, Hough变换, 道路标志线