人脸识别在远程智能监控系统中的研究与实现  

Research and implementation of human face recognition in remote intelligent monitoring system

作　　者：安海平 马行 穆春阳 尹诚 AN Haiping;MA Xing;MU Chunyang;YIN Cheng(Institute of Information and Communication Technology,North Minzu University,Yinchuan 750021,China;College of Mechatronic Engineering,North Minzu University,Yinchuan 750021,China)

机构地区：[1]北方民族大学信息与通信技术研究所,宁夏银川750021 [2]北方民族大学机电工程学院,宁夏银川750021

出　　处：《现代电子技术》2019年第12期176-179,共4页Modern Electronics Technique

基　　金：国家自然科学基金(61163002);宁夏自然科学基金(NZ16086);国家民委中青年英才培养计划(2016GQR10);宁夏高等学校科学研究项目(NGY2016167);北方民族大学重点科研(2015KJ03);北方民族大学研究生创新项目(YCX18069)~~

摘　　要：文中设计一种远程智能视频监控系统,通过加入人脸识别技术,对视频中感兴趣的人脸进行提取并识别。考虑到使用便利性以及成本节约,系统采用嵌入式微处理器平台,并使用Linux作为操作系统。在底层利用V4L2来驱动摄像头输入,利用TCP/IP网络通信协议进行传输。系统采用的是B/S架构,通过浏览器的形式可以查看实时视频监控画面。为了改进普通视频监控的局限性,加入人脸识别技术使其可以自主地筛选出视频中出现人员的信息。采用Adaboost算法检测出人脸,然后通过训练人脸模型库,利用LBP算法对检测出的人脸进行识别。传统的被动视频监控系统存在视频数据利用率不高、资源浪费、增加人工成本等问题。与之相比,基于人脸识别的视频监控系统能够主动地识别视频中人员信息,保存识别的人脸数据,用户可以方便查看视频中出现的人员信息,有着较好的实用性。A remote intelligent video monitoring system is designed in this paper to extract and recognize the interested human faces in the video by adding the human face recognition technology. Taking the usage convenience and cost savings into account,the embedded microprocessor platform and operating system Linux are adopted for the system. In the bottom layer,the V4L2 is used to drive the camera input,and the TCP/IP network communication protocol is used for transmission. The B/S architecture is adopted for the system. The real-time video monitoring screen can be viewed in the form of browser. The human face recognition technology is added to make it possible to autonomously filter the personnel information appearing in the video,so as to improve the limitations of ordinary video monitoring. The human faces are detected by using the Adaboost algorithm,and then recognized by training the human face model library and using the LBP algorithm. In comparison with the traditional passive video monitoring system which has the problems of low utilization rate of video data,waste of resources,and increase of labor costs,the video monitoring system based on human face recognition can actively recognize the personnel information in the video,save the recognized human face data,and make users conveniently view the personnel information appearing in the video, which has a good practicality.

关 键 词：人脸识别 智能监控系统 视频监控 V4L2 B/S架构 ADABOOST算法 

分 类 号：TN948.64.34[电子电信—信号与信息处理]