一种基于视频的交通事故检测方法  

A Video-based Method of Traffic Accident Detection

作　　者：拜佩[1] 李金屏[1] 

机构地区：[1]济南大学信息科学与工程学院,山东济南250022

出　　处：《济南大学学报（自然科学版）》2012年第3期282-286,共5页Journal of University of Jinan(Science and Technology)

基　　金：国家自然科学基金(60773109)

摘　　要：针对目前交通事故检测中存在因疲劳等原因造成交通事故漏检的现象,提出一种基于视频的交通事故检测方法。该方法包括3部分:预处理模块,检测车辆和分析车辆状态;交通事故检测模块,判断是否会发生碰撞;报警模块,主要进行事故实时报警且实时存储事发图片。以车流量较小的路口为例进行仿真实验,结果表明,该系统能准确地预测并判断交通事故的发生。For the phenomenon existing in current traffic incident detection, which accidents are undetected due to fatigue, an effec- tive intelligent traffic incident detection method is presented. This method includes three parts：preprocessing, detecting and analyzing vehicle conditions. Traffic detection section is to determine whether collisions occur or not, and alarm part is mainly for real-time alerting and real-time storing accident incident image. Taking intersections of smaller vehicle flows for simulation experiment, this system can ac- curately forecast and determine the incidence of traffic accidents.

关 键 词：目标检测 跟踪 交通事故 碰撞 

分 类 号：TP3-05[自动化与计算机技术—计算机科学与技术]