基于音视频分析的区域安防管控平台

详细信息    查看全文 | 推荐本文 |

英文篇名：Regional Security Management Platform Based on Audio and Video Analysis 作者：蔡烜 ; 蒋龙泉 ; 冯瑞 英文作者：CAI Xuan;JIANG Longquan;FENG Rui;Internet of Things Technology Research and Development Center,The Third Research Institute of Ministry of Public Security;School of Computer Science, Fudan University; 关键词：视频分析 ; 音频分类 ; 深度学习 ; 人脸识别 ; 人群密度估计 ; 异常声音识别 ; 区域安防管控 英文关键词：Video analysis;;Audio classification;;Deep learning;;Crowd density estimation;;Abnormal voice recognition;;Regional security management 中文刊名：WXDY 英文刊名：Microcomputer Applications 机构：公安部第三研究所物联网技术研发中心;复旦大学计算机科学技术学院; 出版日期：2019-06-14 出版单位：微型电脑应用 年：2019 期：v.35;No.314 基金：上海市科委项目(17511101702) 语种：中文; 页：WXDY201906007 页数：4 CN：06 ISSN：31-1634/TP 分类号：21-24

摘要

针对广场等区域人流量大、安保要求高的特点,设计并实现了一套基于音视频识别的区域安防管控平台。该平台从区域安防的需求出发,设计了嫌疑人员报警、人群拥挤报警和异常声音报警三大功能,以人脸识别、人群密度估计、异常声音识别三个音视频识别算法作为支撑,实现对区域内与人相关的异常事件的预警。首先介绍了基于深度神经网络的人脸识别算法,然后介绍了一种引入注意力机制的卷积神经网络模型来实现的人群密度估计算法,和一种基于多卷积神经网络模型融合的异常声音识别算法;最后介绍了平台需求与设计过程,主要包括平台建设的需求分析和界面设计。
        We design and complete a regional security control platform based on audio and video recognition in order to protect people in the area from dangerous situation. The platform is designed three functions: suspect alarm, crowd crowding alarm and abnormal voice alarm. It uses face recognition, crowd density estimation and abnormal voice recognition as the support to realize the early warning of abnormal events in the region. Firstly, it introduces the face recognition algorithm based on deep neural network, then a convolutional neural network model with attention mechanism is introduced to realize the population density estimation algorithm. Finally, it introduces an abnormal voice recognition algorithm based on multi-convolutional neural network model fusion. At last it introduces the platform requirements and design process, including the platform construction requirement analysis and interface design.

引文

[1] 邹国锋,傅桂霞,李海涛等.多姿态人脸识别综述[J].模式识别与人工智能,2015,28(7):613-625.
    [2] 丁莲静,刘光帅,李旭瑞,等.加权信息熵与增强局部二值模式结合的人脸识别[J].计算机应用,2019(4):1-8.
    [3] Arun Kumar Dubey,Vanita Jain.A review of face recognition methods using deep learning network[J].Journal of Information and Optimization Sciences,2019,40(2):547-558.
    [4] Sugiura,Motoaki,Miyauchi,Carlos Makoto,Kotozaki,Yuka,et al.Neural Mechanism for Mirrored Self-face Recognition[J].Cerebral cortex,2015,25(9):2806-2814.
    [5] Hao-xiang Zhang,Peng An,De-xin Zhang.Application of robust face recognition in video surveillance systems[J].Optoelectronics letters,2018,14(2):152-155.
    [6] Youmei Zhang,Chunluan Zhou,Faliang Chang,Alex C.Kot.Multi-resolution attention convolutional neural network for crowd counting[J].Neurocomputing,2019,Volume 329:144-152.
    [7] Luo H,Sang J,Wu W,et al.A High-Density Crowd Counting Method Based on Convolutional Feature Fusion[J].Applied Sciences,2018,8(12):2367.
    [11] Gunduz A E,Temozel T T,Temizel A.Density estimation in crowd videos[C].//2014 22nd Signal Processing and Communications Applications Conference:2014 22nd Signal Processing and Communications Applications Conference (SIU2014),23 – 25 April 2014,Trabzon,Turkey.2014:822-825.
    [12] 曹金梦,倪蓉蓉,杨彪.基于多尺度多任务卷积神经网络的人群计数[J].计算机应用,2019,39(1):199-204.
    [13] Brian M F,Justin S,Pablo B J.Adaptive Pooling Operators for Weakly Labeled Sound Event Detection[J].IEEE/ACM Transactions on Audio,Speech,and Language Processing,2018,26(11):2180-2193.
    [14] 王开武.基于深度神经网络的异常声音事件检测[D].重庆:重庆大学,2018.
    [15] 吴晓东.智能视频监控技术在智慧城市中的深入应用[J].设备管理与维修,2019(6):150-152.
    [16] 董炜.智能视频分析技术在智慧安防中的应用与展望[J].电子技术与软件工程,2019(7):251-252.
    [17] 任龙刚,王伟,刘峰,等.基于平安校园建设的高校安全管理体系构建——以西安欧亚学院为例[J].安全,2019,40(3):68-71.
    [18] 黄凯奇,陈晓棠,康运锋,等.智能视频监控技术综述[J].计算机学报,2015,38(6):1093-1118.