摘要
异常行为的检测在智能监控系统中具有十分重要的应用。论文提出了一种将关键帧与加权方向角熵值结合的异常行为智能检测方法。该方法首先利用一种像素级的背景提取算法(Visual Background Extractor,VIBE)从视频序列中检测出运动目标,然后提取运动目标幅值加权的方向角特征,并利用改进的无监督聚类方法提取关键帧,最后结合幅值加权方向角熵的方法对异常行为进行检测分类。实验结果表明,该算法能够在指定的监控区域内有效地检测人体行为,并能准确地识别出人体异常行为,实现对监控视频的智能分析。
The detection of abnormal behavior is very important in intelligent surveillance system. In this paper,an intelligent detection method for abnormal behavior is proposed,which combines key frames with weighted direction angle entropy. This method firstly uses the ViBe algorithm to detect the moving object from the video sequences,and then the amplitude weighted moving target direction angle characteristics is extracted,and the improved unsupervised clustering method is used for key frame extraction,final. ly the method of entropy weighted amplitude direction angle of abnormal behavior detection and classification is used. The experi. mental results show that the algorithm can detect human behavior effectively in the designated monitoring area,and recognize the human abnormal behavior accurately,and realize intelligent analysis of surveillance video.
作者
王燕妮
雒津津
王殿伟
WANG Yanni;LUO Jinjin;WANG Dianwei(Department of Information and Control Engineering,Xi'an University of Architecture and Technology,Xi'an 710055;School of Communication and Information Control Engineering,Xi'an University of Posts and Telecommunications,Xi'an 710199)
出处
《计算机与数字工程》
2019年第9期2281-2285,共5页
Computer & Digital Engineering
基金
陕西省自然科学基础研究计划项目(编号:2016JM6079)
陕西省教育厅专项科研基金项目(编号:14JM1429)资助
关键词
异常行为检测
幅值加权方向角熵值
关键帧
无监督聚类
abnormal behavior detection
amplitude weighted directional angular entropy
critical frame
unsupervised clus. tering