基于像素分析的轨迹异常车辆的识别法  

Method of identification trajectory abnormal vehicle based on pixel analysis

作　　者：成春晟[1] 

机构地区：[1]南京邮电大学自动化学院

出　　处：《电子测试》2012年第6期29-34,共6页Electronic Test

摘　　要：交通异常情况检测一直是交通管理中的重要任务,其在智能交通系统中显得尤为重要。传统的检测方法首先对目标物体(行人和车辆)进行区分,然后再对提取的车辆进行轨迹异常判断。在车流量日益加剧的今天,此种方法增加了计算机的运算复杂度。针对上述算法计算量过于复杂的问题,本文提出了基于像素点的背景方法,首先结合隐含马尔可夫模型(Hidden Markov Model)和共发模型(Co-occurences model),对视频中的异常像素点进行判断,然后通过仿三维模型车辆进行识别的方法,对由异常像素点组成的车辆进行最终的识别。实验结果表明该算法具有较强的有效性和在智能交通系统中有较好的应用。The traffic abnormity identification is an important task in the traffic manage system, and it becomes very important in the ITS（Intelligent Transport System）. the traditional detection methods first distinguish the target objects （pedestrians and vehicles） and then judge the abnomal vehicle, this method increases the computational complexity as traffic volume grow quickly in modem time. For the computational algorithm is too complex.we take a method based on pixel background, combined with Hidden Markov Model and Cooccurences model., then we use three-dimensional models to identification vehicle..The results of expefments show that this algorithm has availability and good results in the ITS.

关 键 词：背景方法 隐含马尔可夫模型 共发模型 仿三维模型 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]