基于过街行人检测的路口自适应交通信号控制

Adaptive Traffic Signal Control on Urban Intersection by Waiting-to-cross Pedestrian Detection

针对行人密度波动大的路口的交通信号配时优化问题,提出了以行人过街请求为主的交通信号配时,搭建了由视频采集、行人检测和信号配时构成的交通信号控制系统。视频采集单元采集行人的图像数据;行人检测算法是利用背景重构实时构建行人等候区的背景,继而以背景减法检测等候过街的行人;信号配时由等候过街的行人数和等候时长决定。对实际的交通路口进行仿真,结果表明:在行人低峰时段,行人和车辆平均延误分别为5.62 s/veh和15.99 s/ped,在略微增加行人延误基础上,车流的延误大大降低;在行人高峰时段,车辆平均延误为52.13 s/veh,行人平均延误降至8.35 s/ped,优先保证大密度人群顺利通过路口。

To optimize the time distribution from traffic signal at urban intersections with unstable crowd density, an adaptive traffic signal control system based on pedestrian request was developed. This adaptive signal control system consisted of image sequence capture by video, pedestrians＇ detection and signal phase transition modules. Image sequence was captured by sequence capture unit. In pedestrians＇ detection algorithm, a background reconstruction technology was applied to build background of scene, and then the waiting pedestrians were acquired in real-time through background subtraction. The pedestrians＇ green phase was determined based on the number and associated waiting time of pedestrians i.n the waiting area. The effectiveness of algorithms was tested in a real road intersection. The mean delay of pedestrian and vehicle are 5.62 s/veh and 15.99 s/ped in low pedestrians peak. While the pedestrians are in high peak, the mean delay of vehicle is 52.13 s/veh and the mean delay of pedestrian is 8.35 s/ped.