中文摘要：

提出了一种在ROI（region ofinterest）即感兴趣区域中建立搜索窗口的车道识别算法,使其能够很好地处理非理想路况的各种不确定因素,并建立实时动态搜索窗口,除了能精确地追踪车道标志线之外,还能应用于弯路路况.运用LMedSquare曲线拟合算法提高了车道拟合过程的稳健性和精确性.由于ROI的建立,整个识别过程的计算量大大降低,适应实时性要求.

英文摘要：

A lane detection algorithm is brought forward, which set searching windows in region of interest （ROI）. The algorithm can better deal with the uncertainty factor of road surface, and establish real-time dynamic searching windows. Besides tracking lane mark line more accurately, it can also be applied to curved road conditions. The application of least median square （LmedS） curve fitting algorithm improves the robustness and accuracy of mark line fitting. With the introduction of ROI, the computing effort in searching process can be reduced greatly, so meeting the real-time requirements.