???????????????????????о?

黄灯政策问题已成为道路交通安全领域的研究热点.本文开发了基于视频处理技术的交叉口检测系统，自动采集黄灯期间的交通数据，包括黄灯启亮时车辆到停止线距离、速度、加速度和黄灯结束后车辆的通过情况，建立黄灯期间的驾驶员“停车/行进”决策行为模型和“抢黄灯”行为结果模型.利用北京市“闹市口大街—宣武门西大街”和“中关村东路—成府路”的视频数据对模型进行检验，两个模型的预测结果与实际观测误差分别为8.6%和2.5%.实验结果表明，视频检测技术和驾驶员行为模型能够应用于交叉口的控制管理，本文提出的模型与方法可为相应交通管理措施的制定和实施提供技术支持和理论依据.

Driver Behavior during Yellow Interval Based on Video Detection Technology

Yellow interval policy is a hot topic in the field of traffic safety. Traffic parameters are collected automatically by videos detection technology, including the distance to stop-line, speed, acceleration and the passing status of the vehicles. With these data, the“Stop/Go”decision model and the result of“Yellow Interval Running”model are set up. Predictions of the two models with the observed errors are 2.5% and 5.3%. The test results show that the video detection technology and driver behavior models can be applied to signalized intersection control management, which may provide appropriate support for the development and implementation of traffic management measures.