基于面部生成的视频序列多角度人脸识别系统  

MULTI-ANGLE FACE RECOGNITION SYSTEM FOR VIDEO SEQUENCE BASED ON FACE SYNTHESIS

作　　者：张子豪 张华琰 张雷 Zhang Zihao;Zhang Huayan;Zhang Lei(School of Electrical and Information Engineering,Beijing University of Civil Engineering and Architecture,Beijing 100044,China;Shenzhen Institute of Artificial Intelligence and Robotics for Society(AIRS),Shenzhen 518129,Guangdong,China;Beijing Key Laboratory of Robot Bionics and Function Research,Beijing 100044,China)

机构地区：[1]北京建筑大学电气与信息工程学院,北京100044 [2]深圳市人工智能与机器人研究院,广东深圳518129 [3]机器人仿生与功能研究北京市重点实验室,北京100044

出　　处：《计算机应用与软件》2024年第8期210-218,共9页Computer Applications and Software

基　　金：智能机器人与系统高精尖创新中心建设项目(00921917001)。

摘　　要：针对人机交互系统捕捉大姿态侧脸图像过程中面部纹理缺失严重导致的人脸识别效果不佳问题,提出一种基于面部生成的视频序列多角度人脸识别系统。对传统人脸识别系统中的检测和对齐模块进行改进,将生成对抗网络(GAN)与两种主流面部特征点定位方法相结合。通过增加预处理过程实现正面人脸生成,并针对不同侧脸角度设定系统阈值,还原丢失面部特征。实验结果表明,该系统能有效提高人脸定位精度,并最多可将识别准确率提升18.85%。To solve the problem of poor face recognition effect caused by serious lack of facial texture in the process of human-machine interaction system capturing large pose profile images,this paper proposes a multi-angle face recognition system for video sequences based on face synthesis.By improving the detection and alignment module in traditional face recognition system,it combined the generative adversarial networks(GAN)with two mainstream facial landmarks localization methods.The front face was generated by adding the face preprocessing process,and the system threshold was set for different side face angles to restore the lost facial features.The experimental results show that the proposed system can effectively improve the face location accuracy,and the recognition accuracy can be improved by up to 18.85%.

关 键 词：人脸识别 面部生成 生成对抗网络 面部特征点定位 图像处理 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]