Cross-Age Face Recognition Algorithm Based on HOG-CSLBP and Deep Learning

Yu-ping HU

doi:10.13718/j.cnki.xsxb.2020.03.018

Aiming at the problem of low recognition accuracy in face recognition, a cross-age face recognition algorithm based on deep learning has been proposed. In this method innovatively, the combination of Histogram of Oriented Gradient (HOG) and Center Symmetric Local Binary Pattern (CSLBPS) has been conducted for face image feature extraction and the image fusion features been obtained including the structure and intensity information, and then the binary tree been used to feature the features. The dimensionality reduction feature has been used as the visual layer input of the deep belief network, which makes up for the defect that the depth of the new belief network cannot meet the local features of the image. The test samples have been learned through the trained deep network model, and the features been classified and identified at the top level of the deep belief network. The experimental results show that the proposed method can realize face recognition with high precision, and compared with other methods, the performance of this method is better than other methods, which shows the feasibility and effectiveness of the proposed method.

HTML

人脸识别是一个研究热点，其面临的挑战之一是实现时间不变性^[1-3]，随着人脸识别算法的发展，其识别精度也随之增加，其应用领域也越来越广泛，如监控、门禁、金融支付和远程身份识别等^[4-5].尽管不同外界条件下的人脸识别已经取得了很多研究，但是跨年龄人脸识别仍然是一个挑战^[6-7].老化的人脸识别具有许多有效的实际应用，例如寻找失踪儿童，根据照片识别罪犯和多个登记检测问题.

局部二值模式选择是解释数字图像纹理和模型的一种非常强大的方法，因此是人脸识别系统中特征提取的理想选择.首先将面部图像分割成小区域，再提取特征，然后将其连接成单个特征向量.该向量形成面部区域的有效表示，并可用于测量图像之间的相似性^[8].从原始面部图像导出面部表示，是成功的面部表情识别方法的必要步骤.

大多数常规面部识别系统并不适合处理随年龄增长的面部识别，且许多早期的研究都集中在年龄估计方面，只有少数研究与衰老的人脸识别直接相关. Zou等^[9]提出一种跨年龄人脸识别模型，该模型构造了一种按时间顺序排列的参考集来消除输入图像的年龄差异，利用图库集的身份信息进行判别分析，从而可以在参考编码后进一步区分子空间中的人. Li等^[10]提出了一种老化人脸识别方法，其中使用了基于局部模式选择的分层学习模型.在基于两级学习的分层模型的帮助下解决了该问题.在第一阶段，基于局部模式选择从低微结构学习有效特征，从而减少用户内差异.在第二阶段，基于低级结构的输出细化更高级别的信息. Li等^[11]提出深度卷积网络的跨年龄人脸识别方法，该模型具有共享相同特征层的年龄歧视网络的辨别网络，通过交替训练融合网络和组合因子模型，高级间扩展和类内紧凑性有效地分离跨年龄身份特征. Li等^[12]等人提出一种新的距离度量优化驱动的深度学习框架，用于年龄不变的人脸识别，以端到端的方式同时学习特征表示和相似性度量，使用新颖的优化方法和精心设计的培训策略培训联合网络.

针对以上跨年龄人脸识别方法中识别精度低的问题，本文提出一种基于深度信念网络的跨年龄人脸识别方法.该方法将方向梯度直方图(Histogram of Oriented Gridients，HOG)和中心对称局部二值模式(Center Symmetric Local Binary Pattern，CSLBP)方法联合起来，融合提取图像轮廓特征以及纹理特征，然后使用深度信念网络对特征进行训练分类，实验结果表明，对于跨年龄的人脸识别，所提算法能够得到较高的识别精度.

3. 结语

本文提出一种HOG和CSLBP融合特征的深度学习跨年龄人脸识别算法，该算法使用HOG和CSLBP组合用于获得图像的结构和颜色信息，然后将融合特征作为深度信念网络输入，对特征图像进行预学习与微调，最后使用Logistic回归对人脸图像进行分类.实验结果表明，本文方法能够实现跨年龄人脸的识别，且识别精度优于其他算法，说明本文算法行之有效.未来工作是对本文算法进行深入研究，进一步减少本文算法的时间消耗.

Figure (7) Table (2) Reference (12)

Name
	Name cannot be empty!
E-mail
	Mailbox cannot be empty! Mailbox cannot be empty!
Telephone
	Mobile number cannot be empty! Please enter a valid mobile number!
Title

Content
Verification Code

[1]	SCHROFF F, KALENICHENKO D, PHILBIN J.FaceNet: a Unified Embedding for Face Recognition and Clustering[C]//2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).Boston: IEEE, 2015. Google Scholar
[2]	WEN Y D, ZHANG K P, LI Z F, et al.A Discriminative Feature Learning Approach for Deep Face Recognition[M]//Computer Vision-ECCV 2016.Cham: Springer International Publishing, 2016. Google Scholar
[3]	KARCZMAREK P, KIERSZTYN A, PEDRYCZ W, et al.An Application of Chain Code-based Local Descriptor and Its Extension to Face Recognition[J].Pattern Recognition, 2017, 65:26-34. doi: 10.1016/j.patcog.2016.12.008 CrossRef Google Scholar
[4]	嵇新浩.基于Gabor和HOG的新型人脸识别算法[J].西南师范大学学报(自然科学版), 2017, 42(12):96-104. Google Scholar
[5]	潘禹岐, 江铭炎, 张振月.基于流形的局部加权协从表示人脸识别[J].计算机应用研究, 2016, 33(7):2206-2209. doi: 10.3969/j.issn.1001-3695.2016.07.062 CrossRef Google Scholar
[6]	DU L, LING H B.Cross-age Face Verification by Coordinating with Cross-face Age Verification[C]//2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).Boston: IEEE, 2015. Google Scholar
[7]	CHEN B C, CHEN C S, HSU W H.Cross-Age Reference Coding for Age-Invariant Face Recognition and Retrieval[M]//Computer Vision-ECCV 2014.Cham: Springer International Publishing, 2014. Google Scholar
[8]	JUEFEI-XU F, SAVVIDES M.Subspace-Based Discrete Transform Encoded Local Binary Patterns Representations for Robust Periocular Matching on NIST's Face Recognition Grand Challenge[J].IEEE Transactions on Image Processing, 2014, 23(8):3490-3505. doi: 10.1109/TIP.2014.2329460 CrossRef Google Scholar
[9]	ZOU H S, HU H F.Cross-age Face Recognition Using Reference Coding with Kernel Direct Discriminant Analysis[C]//2017 IEEE International Conference on Image Processing (ICIP).Beijing: IEEE, 2017. Google Scholar
[10]	LI Z F, GONG D H, LI X L, et al.Aging Face Recognition:a Hierarchical Learning Model Based on Local Patterns Selection[J].IEEE Transactions on Image Processing, 2016, 25(5):2146-2154. doi: 10.1109/TIP.2016.2535284 CrossRef Google Scholar
[11]	LI H X, HU H F, YIP C.Age-Related Factor Guided Joint Task Modeling Convolutional Neural Network for Cross-Age Face Recognition[J].IEEE Transactions on Information Forensics and Security, 2018, 13(9):2383-2392. doi: 10.1109/TIFS.2018.2819124 CrossRef Google Scholar
[12]	LI Y, WANG G R, NIE L, et al.Distance Metric Optimization Driven Convolutional Neural Network for Age Invariant Face Recognition[J].Pattern Recognition, 2018, 75:51-62. doi: 10.1016/j.patcog.2017.10.015 CrossRef Google Scholar

Message Board

Cross-Age Face Recognition Algorithm Based on HOG-CSLBP and Deep Learning

Abstract

References

Access History

通讯作者: 陈斌, bchen63@163.com

Article Metrics

Access History

Other Articles By Authors