监控场景下视频中全局移动对象的异常行为自动识别  

Monitor the Scenario of Global Moving Objects in Video Automatically Identify Abnormal Behavior

作　　者：郑良仁[1] 靳宗信[1] 蒋文娟[1] 杨勇[1] 

机构地区：[1]黄河科技学院信息工程学院,郑州450063

出　　处：《科学技术与工程》2017年第26期81-85,共5页Science Technology and Engineering

基　　金：国家青年科学基金(61502432);河南省高等学校重点科研项目(17B520021);河南省高等学校重点科研项目(15B520015)资助;郑州市地方高校急(特)需专业建设项目(zzlg201608);郑州市智能图像处理与识别重点实验室项目(郑[2013]2号)

摘　　要：传统异常行为识别方法容易受到外界环境的干扰,导致识别精度低下;且开销较高。为此,提出一种新的监控场景下视频中全局移动对象的异常行为自动识别方法。对异常行为特征进行提取,选用的特征为目标运动轨迹特征和外接矩形框宽高比特征,通过位置动态对监控场景下视频中全局移动对象运动轨迹特征进行提取,通过最小外接矩形框对移动对象进行标记。依据特征提取结果,通过异常测量函数实现监控场景下视频中全局移动对象的异常行为自动识别。实验结果表明,采用所提方法对全局移动对象的异常行为进行自动识别,不仅识别精度高,且开销较低。The traditional method of abnormal behavior recognition is easy to be disturbed by the external environment,which leads to low recognition accuracy and high overhead. Therefore,the method of automatic recognition of abnormal behavior of a global monitoring scene under the new video moving objects,the abnormal behavior feature extraction,feature selection for target motion trajectory feature and rectangle aspect ratio features,based on the global monitoring scene in the video moving object trajectory characteristics of dynamic position extraction are presented. The smallest rectangle marked on moving objects. According to the results of feature extraction,the abnormal behavior of the moving objects in the video is monitored automatically by the anomaly measurement function.The experimental results show that the proposed method can identify the abnormal behavior of global moving objects automatically,and has high accuracy and low cost.

关 键 词：监控场景 视频 全局移动对象 异常行为 自动识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]