摘要
针对目前光线、角度、遮挡等因素影响人脸识别系统识别率以及基于卷积神经网络CNN的人脸检测和识别算法的计算量比较复杂及鲁棒性不强等问题,研究了一种基于CNN的人脸检测和识别算法,并基于IPv6搭建了一套人脸自动识别系统。该系统采用C/S架构,客户端可部署在任意Windows系统中,服务端运行在赛尔网络云平台服务器。其中,客户端完成人脸信息录入、识别信息展示以及数据库管理等功能;服务器端完成人脸检测和识别功能。实验表明,在一定的条件范围内,该系统人脸识别率可以达到IPv6实际应用的需求。
In view of the fact that the face recognition system is susceptible to sight、angle、occlusion,and the computational complexity and robustness of the face detection and recognition algorithm based on Convolutional Neural Network(CNN),a kind of algorithms based on CNN was discussed.Based on IPv6,an automatic face recognition system was developed.This system adopts C/S architecture,the client can be deployed in any Windows system,and the server runs on CERNET cloud platform server.Among them,the client completes functions such as face information entry,identification information display,and database management;the server is responsible for face detection and recognition functions.Experiments show that the face recognition rate of this system can meet the needs of IPv6 practical applications within a certain range of conditions.
作者
杨静
王艳花
郭亚勤
史健芳
YANG Jing;WANG Yanhua;GUO Yaqin;SHI Jianfang(Information Center,Taigyuan Uniersity of Technology,Taiyuan 030024,China;Information and Communication Branch of State Grid Shanxi Electric Pover Company,Taiyuan 030000,China;Instiute of Information and Computer,Taiyuan Unirersity of Technology,Jinzhong 030600,China)
出处
《电子设计工程》
2021年第12期143-147,共5页
Electronic Design Engineering
基金
赛尔网络下一代互联网技术创新项目(NGII20170615)。
