基于机器视觉与信息共享的交叉路口交通安全预警

摘要

为提高道路交叉路口车辆及行人的通行安全性,提出了一种基于机器视觉与信息共享的全局域内的人车安全预警系统.该系统分为路侧设备和车载设备,通过路侧摄像机获取全路口实时视频图像,应用机器视觉技术解算车辆和行人的运动状态,通过计算碰撞时间预测监控目标继续通行的安全程度,路侧设备向危险车辆精准传递报警信息.预警系统向危险行驶车辆精准传递报警信息.对预警系统中的目标检测及运动分析进行了仿真和测试.试验表明:该预警方案在车辆检测、运动分析和安全预测方面在给定测试条件下具有可行性.%A safety early warning system of vehicles and vulnerable road users in the global domain was proposed based on machine vision and information sharing to improve the traffic safety at road intersections. The system consisted of roadside unit and on board unit. The roadside unit cameras captured intersection real time videos. The machine vision technology calculated the motion states of vehicles and vulnerable road users. The traffic safety degrees of the monitoring targets were judged by calculating the time to collision. The roadside unit released alarm information precisely to the dangerous vehicle. The system transmitted accurately warning information to the dangerous vehicle. The target detection and the motion analysis were simulated and tested for the system. The test results show that the early warning scheme is feasible in vehicle detection, motion analysis and safety prediction under the test conditions.