双摄像头环境下交叉口控制延误提取的误差统计

Error Statistics of Signalized Intersection Control Delay Detect Using Two Cameras

对双摄像头环境下控制延误提取的过程进行了介绍,主要对控制延误提取的误差指标进行了验证。以曹安路-绿苑路交叉口西进口道为参数提取的场景,首先介绍了控制延误参数提取的原理;然后对获取的单个车辆的行程时间值与真实的行程时间值进行对比,使用平均误差指标得到系统提取的单辆车进行比较,得到行程时间值的误差值接近0,使用塞尔指标验证了课题研究中所提取的行程时间与实际的行程时间,结果表明可以认为检测得到的值与真实值相同。之后通过相关的处理,获取按周期统计间隔得到的车均信控延误值,考虑所获取的车均延误值的大小,选用平均相对误差指标作为周期延误参数误差统计的指标,使用k-s验证了延误参数的平均相对误差指标符合正态分布,最后使用t检验验证了课题研究中所提取的周期控制延误平均相对误差指标的误差值小于10%,从而证明了参数提取方法的有效性。

This paper describes the procedure of the signalized intersection control delay detect using two cameras and analyses the parameter＇s error index in detail. Using the Caoan-Lvyuan intersection＇s west approach scene, the theory of control delay collecting is introduced firstly. Secondly, the mean error index is used to compare each vehicle＇s detected travel time and the real travel time. For the mean error is nearly 0, the Theil＇s inequality coefficient is used to test the hypothesis that the detect travel time is equal to the real travel time. Through some processing, the average vehicle control delay is obtained for 1 cycle. Considering the delay＇s value, the mean relative error is used for the error＇s index. The statistics method is used to test that the control delay＇s mean relative error is no more than 10% , which proves the method＇s effectiveness.