摔倒行为视频监控及报警系统  

Video monitoring and alarm system for fall behavior

作　　者：赵琴 赵团结 郑新桥 秦琴 龙念 邓超 ZHAO Qin;ZHAO Tuanjie;ZHENG Xinqiao;QIN Qin;LONG Nian;DENG Chao(Internation Education College,Wuchang Institute of Technology,Wuhan 430065,China;College of Automobile and Traffic Engineering,Wuhan University of Science and Technology,Wuhan 430065,China;Intelligent Automobile Engineering Research Institute,Wuhan University of Science and Technology,Wuhan 430065,China)

机构地区：[1]武昌工学院国际教育学院,武汉430065 [2]武汉科技大学汽车与交通工程学院,武汉430065 [3]武汉科技大学智能汽车工程研究院,武汉430065

出　　处：《智能计算机与应用》2024年第2期114-117,共4页Intelligent Computer and Applications

基　　金：武昌工学院科学研究项目(2022KY24);交通行业重点实验室开放课题(JTZL2205);四川省无人系统智能感知控制技术工程实验室开放课题(WRXT2022-001);教育部产学合作协同育人项目(230803612214746);云基物联网高速公路建养设备智能化实验室开放课题(KF_2022_301002)。

摘　　要：针对当前视频监控系统对摔倒行为识别率低的情况,本文提出了一种摔倒行为实时监控预警的方法,解决了视频监控系统对摔倒行为误报、漏报等问题。首先,采用图像识别的方法对人的动作方向特征进行提取;其次,基于运动方向的隐马尔科夫模型(HMM)进行动作分割,在视频中分割出人的一些常见的动作;最后,依据这些动作的组合判断下一时刻是否存在摔倒行为,进而发出警报。该监控预警方法可对进入监控区的人与其他物体域进行自动区分,为行为监控系统提供一种理论依据。This paper proposes a method for real-time monitoring and early warning of fall behavior,addressing issues such as false alarms and missed detections in current video surveillance systems.Initially,image recognition techniques are employed to extract features related to the movement direction of individuals.Subsequently,a Hidden Markov Model(HMM)based on motion direction is used for action segmentation,isolating common movements of people in the video.Finally,based on the combination of these movements,the system predicts potential fall behavior in the next moment and triggers an alarm.This monitoring and early warning method can automatically distinguish between people and other objects entering the monitored area,providing a theoretical basis for behavioral monitoring systems.

关 键 词：摔倒行为 监控系统 图像识别 隐马尔科夫模型 

分 类 号：TN948.6[电子电信—信号与信息处理]