基于深度学习与特征融合的人脸识别算法

doi:10.16180/j.cnki.issn1007-7820.2020.04.004

电子科技 ›› 2020, Vol. 33 ›› Issue (4): 18-22.doi: 10.16180/j.cnki.issn1007-7820.2020.04.004

基于深度学习与特征融合的人脸识别算法

司琴,李菲菲,陈虬

上海理工大学光电信息与计算机工程学院,上海 200093

收稿日期:2019-03-01 出版日期:2020-04-15 发布日期:2020-04-23
作者简介:司琴(1996-),女,硕士研究生。研究方向:模式识别与机器视觉|李菲菲(1970 -),女,博士,教授。研究方向:多媒体信息处理、图像处理与模式识别、信息检索等。|陈虬(1972-),男,博士,教授,博士生导师。研究方向:图像处理与模式识别,计算机视觉,信息检索等。
基金资助:
上海市高校特聘教授(东方学者)岗位计划(ES2015XX)

Face Recognition Algorithm Based on Deep Learning and Feature Fusion

SI Qin,LI Feifei,CHEN Qiu

School of Optical-Electrical and Computer Engineering,University of Shanghai for Science and Technology,Shanghai 200093,China

Received:2019-03-01 Online:2020-04-15 Published:2020-04-23
Supported by:
Program for Professor of Special Appointment (Eastern Scholar) at Shanghai Institutions of Higher Learning(ES2015XX)

摘要/Abstract

摘要：

卷积神经网络在人脸识别研究上有较好的效果,但是其提取的人脸特征忽略了人脸的局部结构特征。针对此问题,文中提出一种基于深度学习与特征融合的人脸识别方法。该算法将局部二值模式信息与原图信息相结合作为SDFVGG网络的输入,使得提取的人脸特征更加丰富且更具表征能力。其中,SDFVGG网络是将VGG网络进行深浅特征相融合后的网络。在CAS-PEAL-R1人脸数据库上的实验表明,将网络深浅特征相融合与在卷积神经网络中加入LBP图像信息与原图信息相融合的特征信息对于提高人脸识别准确率非常有效,可得到优于传统算法和一般卷积神经网络的最高98.58%人脸识别率。

关键词: 特征提取, 特征融合, 卷积神经网络, SDFVGG, 局部二值模式, 人脸识别

Abstract:

The convolutional neural network performed well on face recognition, but the extracted facial feature ignores the local structure features of the face. In order to address this problem, a novel method was proposed, which was based on deep learning and feature fusion. This algorithm made the extracted facial features more comprehensive by using a combination of local binary pattern information and original image information as the input of SDFVGG network, which was a neuralnetwork fusing shallow and deep features of VGG network. Experimental results on the CAS-PEAL-R1 face database demonstrated that the proposed algorithm was very effective for improving the accuracy of face recognition, and achieved a maximum face recognition rate of 98.58% which was better than traditional algorithms and general convolutional neural networks.

Key words: feature extraction, feature fusion, convolutional neural network, SDFVGG, local binary pattern, face recognition

中图分类号:

TP391

司琴,李菲菲,陈虬. 基于深度学习与特征融合的人脸识别算法[J]. 电子科技, 2020, 33(4): 18-22.

SI Qin,LI Feifei,CHEN Qiu. Face Recognition Algorithm Based on Deep Learning and Feature Fusion[J]. Electronic Science and Technology, 2020, 33(4): 18-22.

图/表 9

图 1

图2

图3

图4

图5

图6

表1

表2

表3

参考文献 16

[1]	吴学谦, 李菲菲, 颜艳 , 等. 基于局部二值模式及马尔科夫稳态特征的人脸识别[J]. 电子科技, 2018,31(8):18-20,24.
	Wu Xueqian, Li Feifei, Yan Yan , et al. Face recognition algorithm using LBP and MSF features[J]. Electronic Science and Technology, 2018,31(8):18-20,24.
[2]	Lihong Z, Fei L, Yongjun W . Face recognition based on LBP and genetic algorithm[C]. Yinchuan:2016 Chinese Control and Decision Conference (CCDC), 2016.
[3]	Zhang J, Xiao X . Face recognition algorithm based on multi-layer weighted LBP[C]. Hangzhou:8^th International Symposium on Computational Intelligence and Design (ISCID), 2015.
[4]	Bengio Y, Delalleau O . Onthe expressive power of deep architectures[C]. Espoo:Discovery Science 14^th International Conference , 2011.
[5]	Wang M, Wang Z, Li J . Deep convolutional neural network applies to face recognition in small and medium databases[C]. Hangzhou:4^th International Conference on Systems and Informatics (ICSAI), 2017.
[6]	Lecun Y, Bengio Y, Hinton G . Deep learning[J]. Nature, 2015,521(7553):436-446. doi: 10.1038/nature14539 pmid: 26017442
[7]	Syaffeza A R, Khalil-Hani M, Liew S S , et al. Convolutional neural network for face recognition with pose and illumination variation[J]. International Journal of Engineering & Technology, 2014,6(1):44-57.
[8]	Yan K, Huang S, Song Y , et al. Face recognition based on convolution neural network[C]. Dalian:Chinese Control Conference, 2017.
[9]	Gu S, Ding L . A complex-valued VGG network based deep learing algorithm for image recognition[C]. Wanzhou: Ninth International Conference on Intelligent Control and Information Processing (ICICIP), 2018.
[10]	Russakovsky O, Deng J, Su H , et al. ImageNetlarge scale visual recognition challenge[J]. International Journal of Computer Vision, 2014,115(3):211-252. doi: 10.1007/s11263-015-0816-y
[11]	Dahl G E, Sainath T N, Hinton G E . Improving deep neural networks for LVCSR using rectified linear units and dropout[C]. Vancouver:IEEE International Conference onAcoustics, 2013.
[12]	Gao W, Member S, Cao B , et al. The CASPEAL large-scale chinese face database and baseline evaluations[J]. IEEE Trans on System Man & Cybernetics, 2008,38(1):149-161.
[13]	Xie S, Shan S, Chen X , et al. Learned local Gabor patterns for face representation and recognition[J]. Signal Processing, 2009,89(12):2333-2344. doi: 10.1109/TPAMI.2013.112 pmid: 24356350
[14]	Zhang B, Shan S, Chen X , et al. Histogramof Gabor Phase Patterns (HGPP):a novel object representation approach for face recognition[J]. IEEE Transactions on Image Processing, 2007,16(1):57-68. doi: 10.1109/tip.2006.884956 pmid: 17283765
[15]	Farajzadeh N, Faez K, Pan G . Study on the performance of moments as invariant descriptors for practical face recognition systems[J]. IET Computer Vision, 2010,4(4):272-285. doi: 10.1049/iet-cvi.2009.0140
[16]	Tan H, Yang B, Ma Z . Face recognition based on the fusion of global and local HOG features of face images[J]. IET Computer Vision, 2014,8(3):224-234. doi: 10.1049/iet-cvi.2012.0302

网络	识别率/%
网络	PE	PA	PL
VGG	99.69	97.66	98.44
SDFVGG	99.69	98.18	99.22

网络输入	识别率/%
网络输入	PE	PA	PL
原图	99.69	98.18	99.22
LBP	99.06	96.61	97.66
原图+LBP	98.58	97.12	98.05

不同方法	识别率/%
不同方法	PE	PA	PL
GPCA+LDA^[13] LGBPHS^[14] PZM-RBFNN^[15] HGPP^[14] LLGP^[13] FHOGC^[16] Ours	90.60 95.20 84.80 96.40 98.00 94.90 98.58	82.80 86.80 93.40 91.90 92.00 90.30 97.12	44.80 51.00 63.40 61.70 55.00 68.70 98.05

基于深度学习与特征融合的人脸识别算法

Face Recognition Algorithm Based on Deep Learning and Feature Fusion

RichHTML

PDF (PC)

赞

可视化

摘要/Abstract

引用本文

使用本文

图/表 9

参考文献 16

相关文章 15

Metrics

本文评价

推荐阅读 10

[1]	伍伟嘉,杨俭,袁天辰,邵志慧. 基于支持向量机的轨道结构病害识别研究[J]. 电子科技, 2022, 35(2): 27-33.
[2]	李辉,王一丞. 基于CNNCIFG-Attention模型的文本情感分类[J]. 电子科技, 2022, 35(2): 46-51.
[3]	邵志慧,杨俭,袁天辰,伍伟嘉. 基于排列熵和支持向量机的轨枕病害诊断[J]. 电子科技, 2022, 35(2): 52-58.
[4]	程晓雅,张雷. 基于深度学习的有遮挡人脸识别方法研究[J]. 电子科技, 2022, 35(1): 35-39.
[5]	赵崇,迟蒙蒙,储聪,张鹏. 导盲犬行走机构运动仿真及其视觉识别算法研究[J]. 电子科技, 2021, 34(9): 66-72.
[6]	战荫伟,朱百万,杨卓. 一种车脸识别算法的研究与应用[J]. 电子科技, 2021, 34(8): 1-7.
[7]	廖璐明,张伟. 基于改进VGG16网络的混合批量训练交通标志识别[J]. 电子科技, 2021, 34(8): 8-13.
[8]	司明明,陈玮,胡春燕,尹钟. 融合Resnet50和U-Net的眼底彩色血管图像分割[J]. 电子科技, 2021, 34(8): 19-24.
[9]	马立新,豆晨飞,宋晨灿,杨天笑. 基于特征融合卷积神经网络的绝缘子无损检测[J]. 电子科技, 2021, 34(7): 26-30.
[10]	顾伟,李菲菲,陈虬. 基于多特征融合的行人检测方法[J]. 电子科技, 2021, 34(5): 29-34.
[11]	刘姝,邵杰,张颐婷,张善章. 基于双谱和谱特征的心电信号分类方法[J]. 电子科技, 2021, 34(5): 42-46.
[12]	宋章明,贺慧勇,黄跃俊. 扁平漆包线表面缺陷在线检测研究[J]. 电子科技, 2021, 34(5): 72-78.
[13]	薛永杰,巨志勇. 基于改进AlexNet的鱼类识别算法[J]. 电子科技, 2021, 34(4): 12-17.
[14]	闫书豪,乔美英. 基于一维WConv-BiLSTM的轴承故障诊断算法[J]. 电子科技, 2021, 34(4): 75-82.
[15]	葛靖,刘子龙. 基于CNN和LSTM的睡眠呼吸暂停检测算法[J]. 电子科技, 2021, 34(2): 21-26.