一种基于灰度直方图的交通检测系统  

A Grey-histogram Based Traffic Detection System

作　　者：耿凯[1] 姚丹亚[1] 张毅[1] 

机构地区：[1]清华大学自动化系,北京100084

出　　处：《计算机工程与应用》2006年第4期222-225,共4页Computer Engineering and Applications

基　　金：国家自然科学基金资助项目(编号:60374059)

摘　　要：在中国由于行人和机动车混行的情况非常普遍,因此行人的检测和机动车一样也是交通检测的重要方面。然而国内外在行人与机动车混合交通流检测方面的研究却不多见。文章提出并阐述了一套基于视频的综合交通检测系统。该系统通过视频技术实时检测获得路段或路口上机动车、行人的综合交通数据。系统运用了实用性很强自适应背景差减法进行前景图像分割,并提出了基于灰度直方图的目标跟踪方法,在一定程度上解决了在人车混行中比较严重的遮挡问题。实验结果表明,系统在前景图像提取以及目标跟踪分类等环节都取得了令人满意的成果。The detection of bicycles and pedestrians is as important as the detection of vehicles especially in China.In this paper,a video based traffic detection system is proposed,which can obtain real time traffic data of vehicle,bicycles and pedestrians.So far,it＇s a common problem in multiple objects tracking to resolve the identity of each tracked object after an occlusion.In our system,a self-adaptive background subtraction method is adopted and an object tracking method based on grey-histogram is proposed.And experimental results show that the system resolves the problems due to occlusion effectively at a certain extent.

关 键 词：背景差减法 MBR 灰度直方图 目标跟踪 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]