Multi-view Gait Recognition Method Based on View Transformation

doi:10.19678/j.issn.1000-3428.0057899

Abstract

Abstract: The existing gait recognition methods are limited by multiple factors, including the changes of view, small gait sample size, and under-utilization of the temporal information of gaits.To address the problems and improve the gait recognition performance, this paper proposes a gait recognition method based on view transformation.Through VTM-GAN, Gait Energy Images(GEI) and Chrono Gait Images(CGI) with temporal gait information of different views are mapped to the side view that contains the most abundant gait information in order to break the limitation of views in gait recognition.On the basis of view transformation, the positive and negative sample pairs of gait data in the side view are constructed to extend the volume of network training data.The Spatial-temporal double flow convolutional neural network based on distance measurement is taken as the gait recognition network.Experimental results on the CASIA-B dataset show that the average recognition accuracy of this method in all states and views reaches 92.5%, higher than that of 3DCNN, SST-MSCI and other gait recognition methods.

Key words: gait recognition, view transformation, VTM-GAN network, spatial-temporal double flow Convolutional Neural Network(CNN), CASIA-B dataset

摘要： 针对步态识别中步态视角变化、步态数据样本量少及较少利用步态时间信息等问题，提出一种基于视角转换的步态识别方法。通过VTM-GAN网络，将不同视角下的步态能量图及含有步态时间信息的彩色步态能量图，统一映射到保留步态信息最丰富的侧视图视角，以此突破步态识别中多视角的限制，在视角转换的基础上，通过构建侧视图下的步态正负样本对来扩充用于网络训练的数据，并采用基于距离度量的时空双流卷积神经网络作为步态识别网络。在CASIA-B数据集上的实验结果表明，该方法在各状态、各角度下的平均识别准确率达到92.5%，优于3DCNN、SST-MSCI等步态识别方法。

关键词: 步态识别, 视角转换, VTM-GAN网络, 时空双流卷积神经网络, CASIA-B数据集

CLC Number:

TP391

QU Binjie, SUN Shaoyuan, Samah A. F. Manssor, ZHAO Guoshun. Multi-view Gait Recognition Method Based on View Transformation[J]. Computer Engineering, 2021, 47(6): 210-216.

瞿斌杰, 孙韶媛, Samah A. F. Manssor, 赵国顺. 基于视角转换的多视角步态识别方法[J]. 计算机工程, 2021, 47(6): 210-216.

/ / Recommend / Download Citations

URL: http://www.ecice06.com/EN/10.19678/j.issn.1000-3428.0057899

http://www.ecice06.com/EN/Y2021/V47/I6/210

Figures/Tables 14

References

[1] WANG Kejun,DING Xinnan,XING Xianglei,et al.A survey of multi-view gait recognition[J].Acta Automatica Sinica,2019,45(5):841-852.(in Chinese)王科俊,丁欣楠,邢向磊,等.多视角步态识别综述[J].自动化学报,2019,45(5):841-852.
[2] JOHNSON A Y,BOBICK A F.A multi-view method for gait recognition using static body parameters[C]//Proceedings of the 3rd International Conference on Audio-and Video-Based Biometric Person Authentication.Halmstad,Sweden:[s.n.],2001:301-311.
[3] ARIYANTO G,NIXON M S.Model-based 3D gait biometrics[C]//Proceedings of International Joint Conference on Biometrics.Washington,D.C.,USA:IEEE Press,2011:1-7.
[4] HAN J,BHANU B.Individual recognition using gait energy image[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2006,28(2):316-322.
[5] BOBICK A F,DAVIS J W.The recognition of human movement using temporal templates[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2001,23(3):257-267.
[6] LAM T H W,LEE R S T.A new representation for human gait recognition:motion silhouettes image[C]//Proceedings of International Conference on Biometrics.Hong Kong,China:[s.n.],2006:612-618.
[7] LAM T H W,CHEUNG K H,LIU J N K.Gait flow image:a silhouette-based gait representation for human identification[J].Pattern Recognition,2011,44(4):973-987.
[8] HE Yiwei,ZHANG Junping.Deep learning for gait recognition:a survey[J].Pattern Recognition and Artificial Intelligence,2018,31(5):442-452.(in Chinese)何逸炜,张军平.步态识别的深度学习:综述[J].模式识别与人工智能,2018,31(5):442-452.
[9] LIU Nini,LU Jiwen,TAN Yaping.Joint subspace learning for view-invariant gait recognition[J].IEEE Signal Processing Letters,2011,18(7):431-434.
[10] KUSAKUNNIRAN W,WU Q,ZHANG J,et al.A new view-invariant feature for cross-view gait recognition[J].IEEE Transactions on Information Forensics and Security,2013,8(10):1642-1653.
[11] MAKIHARA Y,SAGAWA R,MUKAIGAWA Y,et al.Gait recognition using a view transformation model in the frequency domain[C]//Proceedings of European Conference on Computer Vision.Berlin,Germany:Springer,2006:151-163.
[12] KUSAKUNNIRAN W,WU Q,LI H D,et al.Multiple views gait recognition using view transformation model based on optimized gait energy image[C]//Proceedings of the 12th IEEE International Conference on Computer Vision.Washington D.C.,USA:IEEE Press,2009:1058-1064.
[13] GODBEHERE A B,MATSUKAWA A,GOLDBERG K.Visual tracking of human visitors under variable-lighting conditions for a responsive audio art installation[C]//Proceedings of the American Control Conference.Montreal,USA:IEEE Press,2012:4305-4312.
[14] RAFAEL C G,RICHARD E W.Digital image processing[M].Englewood Cliffs,USA:Prentice-Hall,Inc.,2007.
[15] WANG Chen,ZHANG Junping,WANG Liang,et al.Human identification using temporal information preserving gait template[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2012,34(11):2164-2176.
[16] ZHU J Y,PARK T,ISOLA P,et al.Unpaired image-to-image translation using cycle-consistent adversarial networks[EB/OL].[2020-02-20].https://arxiv.org/abs/1703.10593.
[17] YU Shiqi,TAN Daoling,TAN Tieniu.A framework for evaluating the effect of view angle,clothing and carrying condition on gait recognition[C]//Proceedings of IEEE International Conference on Pattern Recognition.Washington D.C.,USA:IEEE,2006:441-444.
[18] WU Zifeng,HUANG Yongzhen,WANG Liang,et al.A comprehensive study on cross-view gait based human identification with deep CNNs[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2017,39(2):209-226.
[19] ZENG Wei,WANG Cong.View-invariant gait recognition via deterministic learning[J].Neurocomputing,2016,175(29):324-335.
[20] YU S Q,CHEN H F,REYES E B G,et al.GaitGAN:invariant gait feature extraction using generative adversarial networks[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition Workshops.Washington D.C.,USA:IEEE Press,2017:532-539.

Please choose a citation manager

Content to export