基于全方位视觉传感器的车辆违章检测系统的设计

Design of a Traffic Violation Detection System Based on Omni-Directional Vision Sensors

探讨了利用全方位视觉传感器(ODVS)以及计算机视觉等技术来实现对交岔路口的车辆违章智能视频监控,采用ODVS来获取整个交岔路口的全景视频图像;通过混合高斯模型提取在全景视频图像中的车辆对象前景;使用camshift跟踪算法跟踪行驶车辆对象;依据交通法规,通过对比红绿灯、车道和车辆运动等状态来判断车辆对象是否有违章行为;实验结果表明,所设计的违章检测系统能自动检测出多种车辆违章行为,为交通执法部门提供了一种智能化的检测手段。

A new intelligent monitoring system for detecting traffic violation is presented, the system is based on omni-direetional vision sensor （ODVS） and computer vision technology. In course of omni-directional image acquisition on intersections, ODVS is used. In course of vehicle object foreground extraction, color space conversion and mulgaussian model is utilized. In course of vehicle object tracking, camshift tracking algorithm is proposed. Finally, based on traffic law, traffic lights, lanes and vehicle movement, and other state are compared to estimate vehicles peccancy. The experimental results demonstrate that the proposed scheme can automatically detect variety of vehicles illegal acts, and provide transportation department a intelligent method to detect vehicle violations.