Gait Recognition with Clothing and Carrying Variations Based on GEI and CAPDS Features

Gait recognition is a promising technology in biometrics. The accuracy of gait recognition can be decreased by many interference variations, such as view angle, clothing and carrying. A novel method is proposed based on the Gait Energy Image (GEI) feature and Coordinate-Angle-Position-Distance Skeleton (CAPDS) feature to eliminate the interference of clothing and carrying variations. GEI is a common feature widely used in gait recognition, but it is sensitive to the change of clothing and carrying. The CAPDS proposed in this paper is robust to the clothing and carrying variations. They are fused in backward to complement each other for recognition. Two novel networks, the Paird ResNet (PRN) and the Temporal-Spatial Paired Network (TSPN), are designed to extract the deep features of GEI and CAPDS. The experiments evaluated on the dataset CASIA-B show that the proposed method based on the backward fusion strategy of GEI and CAPDS features can achieve better performance than most methods in gait recognition with clothing and carrying variations.

[1]  Xiaogang Wang,et al.  A Comprehensive Study on Cross-View Gait Based Human Identification with Deep CNNs , 2017, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[2]  Yasushi Makihara,et al.  Gait Energy Response Functions for Gait Recognition against Various Clothing and Carrying Status , 2018 .

[3]  Ausif Mahmood,et al.  Improved Gait recognition based on specialized deep convolutional neural networks , 2015, 2015 IEEE Applied Imagery Pattern Recognition Workshop (AIPR).

[4]  Shiqi Yu,et al.  Pose-Based Temporal-Spatial Network (PTSN) for Gait Recognition with Carrying and Clothing Variations , 2017, CCBR.

[5]  Jian Weng,et al.  Multi-Gait Recognition Based on Attribute Discovery , 2018, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[6]  Bir Bhanu,et al.  Individual recognition using gait energy image , 2006, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[7]  LinLin Shen,et al.  Invariant feature extraction for gait recognition using only one uniform model , 2017, Neurocomputing.

[8]  Xiang Li,et al.  Gait Energy Response Function for Clothing-Invariant Gait Recognition , 2016, ACCV.

[9]  Shilin Wang,et al.  3D human action recognition based on the Spatial-Temporal Moving Skeleton Descriptor , 2017, 2017 IEEE International Conference on Multimedia and Expo (ICME).

[10]  Yasushi Makihara,et al.  GEINet: View-invariant gait recognition using a convolutional neural network , 2016, 2016 International Conference on Biometrics (ICB).

[11]  Shiqi Yu,et al.  GaitGAN: Invariant Gait Feature Extraction Using Generative Adversarial Networks , 2017, 2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW).

[12]  Yueting Zhuang,et al.  Fusing Geometric Features for Skeleton-Based Action Recognition Using Multilayer LSTM Networks , 2018, IEEE Transactions on Multimedia.

[13]  Shiqi Yu,et al.  GaitGANv2: Invariant gait feature extraction using generative adversarial networks , 2019, Pattern Recognit..

[14]  Yaser Sheikh,et al.  OpenPose: Realtime Multi-Person 2D Pose Estimation Using Part Affinity Fields , 2018, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[15]  Jian Sun,et al.  Deep Residual Learning for Image Recognition , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[16]  Dinesh Kumar Vishwakarma,et al.  Covariate Conscious Approach for Gait Recognition Based Upon Zernike Moment Invariants , 2016, IEEE Transactions on Cognitive and Developmental Systems.