道路边界识别是基于机器视觉的智能车辆关键技术之一，提出了一种基于直线道路模型的道路边界识别和跟踪算法，该算法包括道路边界初始识别算法和跟踪算法两部分。在初始道路边界识别阶段，对预处理后的图像进行逐行搜索道路边界候选点，并结合相关的道路约束条件，采用Hough变换对候选边界点进行拟合，以提高鲁棒性。该方法的创新点是在道路跟踪算法中，建立目标搜索区域（OSA），OSA的宽度根据道路边界识别的拟合可信度的变换进行动态调整，并采用随机Hough变换的算法对OSA中的边界点进行拟合，这样显著减少了运算时间和存储空间。试验结果表明，该方法不仅满足道路边界识别的实时性要求，还具有较好的鲁棒性。

The road edge identification is one of the key technology of the intelligent vehicle based on machine vision. A road edge identification and tracking algorithm based on linear road model is developed in this paper. This algorithm includes two stages, one is initial road identification algorithm and the other is road tracking algorithm. Pretreat the image at first at initial road edge identification stages, and then scan the image per-line in order to search the possible road edge points of every row. The method of Hough transform (HT) is used to match these points according to some road constrains. The new idea of this method is to set up the object search area (OSA) firstly at the road tracking stage, and the width of OSA is changing according to the change of the road edge identification reliability. And the possible road edge points are matched by Randomized Hough transform(RHT) method, This algorithm can decrease the algorithm running time and storage space evidently. Experiments prove that the algorithm is robust and fulfills the requirement real time of road edge identification.