基于视频车辆运动轨迹场的交通事件检测方法*

针对现有交通事件检测算法存在的模型复杂、运算量大,并且检测结果滞后的缺点,提出了一种基于视频车辆运动轨迹场的交通事件检测方法.该方法以基于视频的车辆跟踪轨迹为基础,形成正常车辆运动轨迹场,进而将当前车辆轨迹矢量与正常轨迹场矢量比较,最终实现了对超速、慢行、变道和逆行的异常交通事件的检测.研究结果表明,与传统方法相比,该方法算法简单,并且可以快速、准确地对异常交通事件进行检测.

A method of detecting traffic incident

Currently the existing traffic incident detection algorithm has some disadvantages such as complex model ,heavy computation.And the detection result has time lag. In this paper,a method of detecting traffic incident based on vehicle trajectory field is introduced.The method obtains the video-based vehicle tracking trajectory ,establishes a normal vehicle trajectory field,then compares the current vehicle trajectory with the normal vehicle trajectory field and realizes the function of abnormal traffic incident detection such as over-speed,low running speed,changing lane,retrograde in the end.The results show that compared with the traditional approaches,this approach has simple algorithm and,also can detect traffic incidents quickly and accurately.