摘要
近年来,随着各行业对安全认证和监控系统的需求激增,如何准确识别身份信息已经成为了学术界的热点研究方向。生物学信息识别技术由于个体特征的唯一性在识别的准确度上有着得天独厚的优势而迅速崛起。基于深度学习和特征融合理论提出一种人脸识别算法。首先,分析了人脸识别的行业发展现状;其次,阐述了深度神经网络的原理并分析了各种模型的特征;再次,提出一种人脸特征融合算法;最后,在实验中,以不同肤色、人种、性别的人脸图像为实验对象,验证了所提出算法在多种条件下的有效性。
In recent years,with the increasing demand for security authentication and monitoring system in various industries,how to accurately identify the identity information has become a hot research direction in academia.Due to the uniqueness of individual characteristics,biological information recognition technology has a unique advantage in the accuracy of recognition and is rising rapidly.This paper proposes a face recognition algorithm based on deep learning and feature fusion theory.Firstly,the development status of face recognition industry is analyzed.Secondly,the principle of deep neural network is expounded and the characteristics of various models are analyzed.Thirdly,a face feature fusion algorithm is proposed.Finally,face images of different skin color,race and gender are taken as experimental objects to verify the effectiveness of the proposed algorithm under various conditions.
作者
郭天伟
齐金山
杨海东
王超
GUO Tianwei;QI Jinshan;YANG Haidong;WANG Chao(Information Statistics Center, Huai’an Second People’s Hospital, Huai’an 223001, China;School of Computer Science and Technology, Huaiyin Teachers College, Huai’an 223001, China;School of Information Science, Henan Polytechnic University, Jiaozuo 454000, China)
出处
《微型电脑应用》
2020年第11期5-8,22,共5页
Microcomputer Applications
基金
国家自然科学基金项目(61501215)
2019市科技局支持项目(HAB201934)
省人才办第五期333(BRA2017245)。
关键词
深度学习
卷积神经网络
特征融合
人脸识别
deep learning
convolutional neural network
feature fusion
face recognition
