Abstract: The range of interest (ROI) of the vehicle detection is defined after the lane has been detected, then according to the characteristics of the vehicle vertical direction, shadow under vehicles and taillight features to lock the front vehicle position, and the feature information of vehicles are converted into vehicle distance information finally. The algorithm proposed in this paper treats differently with day and night, and utilizes the strategy of single and double camera handover to detect the vehicle distance, which makes the system in short distance vehicle distance detecting with excellent performance. Experiments show that, this system designed in the paper, achieves the anticipated function, can accurately detect multi-lane and short distance vehicle distance, and the efficiency of the system also meets the real-time requirement.

Key words: Lane detection, vehicle distance detection, single and double camera

摘要： 在车道线被侦测出的前提下，前方车辆的侦测(Range of Interest，ROI)范围将会依赖这些车道线特征进行限定，随后根据车底阴影、车辆竖直方向上的特征以及车尾灯特征锁定前方车辆的位置，并且这些车辆的特征信息最终会转换成车距的信息。由于文中算法对白天以及夜晚两种不同情况做了相应差异化处理，且采用单双摄像头切换测距的方式，使得系统在中短距离的车距侦测具有优良的性能。实验表明，文章所设计的系统能够准确的侦测出多车道以及前方中短距离车距，系统的运行效率也能够达到实时性的要求，总体上很好的实现了预期的功能。

关键词: 特征点匹配, 误匹配, 迭代次数