人類走路之步態已經被心理學家證明是對於每個人而言是唯一的。因此，它對於應用在身份辨識是一個有用的特徵。然而，使用步態特徵作身份辨識仍然有一些問題。在這篇論文中，我們將重點放在步態受到衣著影響變化之改善。首先，我們利用沒有受到衣著影響變化之步態，學出不受衣著影響變化之步態特性，並且將此特性模組化。我們稱此模組為正常情況模組。藉由這個正常情況模組，我們可以偵測出在步態中有受到衣著影響變化的部分並且除去之。我們將利用正常情況模組進行除去衣著影響變化之過程，稱之為正常情況相減法。從本篇論文的實驗結果中，我們提出的正常情況相減法的確能夠在人類身份辨識之應用中，比其他現有設法解決衣著影響變化之方法還要有更高的辨識精確度。

Human gait has been shown to be unique and could be treated as a useful signature for human identification. Nevertheless there are still some problems when using the gait for human identification. In this thesis, we focus on the improvement of the clothing-variation problem. First, we model the characteristics of human gait which suffers no clothing-variation as the normal case model. Next, we detect and remove the clothing-variation parts of a gait based on this model. The process of removing the clothing-variation using the normal case model is called the normal case subtraction. The experimental results show that our proposed normal case subtraction outperforms other existing approaches of solving the clothing-variation problem in human identification.

致謝 I
中文摘要 II
ABSTRACT III
1. INTRODUCTION 1
2. RELATED WORK 2
2.1 Gait Feature Extraction 2
2.1.1 Model-based Gait Feature Extraction 2
2.1 .2 Appearance-based Gait Feature Extraction 3
2.1 .3 Gait Energy Image (GEI) 3
2.2 Solving the clothing-variation problem 4
2.2.1 Modeling the clothing-variation changes 4
2.2.2 Considering only the Lower Part of Human Body 5
2.2.3 Considering Both the Dynamic and Static Part of Human Body 6
2.3 Motivation 7
3. PROPOSED METHOD 13
3.1 Normal Case Subtraction 13
3.1.1 Bayesian Normal Case Modeling 13
3.1.2 Normal Case Likelihood term 14
3.1.3 Normal Case Prior term 15
3.1.4 Removal of Clothing-Variation Parts 16
3.2 Feature Projection and Classification 17
4. EXPERIMENTAL RESULTS 23
4.1 Database and Setting 23
4.2 Validation of Normal Case Model 24
4.3 Comparison with Existing Methods 25
4.4 Discussion 26
5. CONCLUSION 40
6. REFERENCES 41

[1] L. Wang, H. Ning, T. Tan, and W. Hu, “Fusion of Static and Dynamic Body Biometrics for Gait Recognition,” IEEE Trans. Circuits and Systems for Video Technology, vol. 14, no. 2, pp. 149-158, Feb. 2004.
[2] D. K. Wagg and M. S. Nixon, “Automated Markerless Extraction of Walking People Using Deformable Contour Models,” Comput. Animation Virtual Worlds, vol. 15, no. 3/4, pp. 399-406, Jul. 2004.
[3] I. Bouchrika and M. S. Nixon, “Model-Based Feature Extraction for Gait Analysis and Recognition,” Proceedings of Mirage: Computer Vision / Computer Graphics Collaboration Techniques and Applications, pp. 150-160, 2007.
[4] D. Kim, D. Kim, and J. Paik, “Gait Recognition Using Active Shape Model and Motion Prediction,” IET Computer Vision, vol. 4, no. 1, pp. 25-36, 2010.
[5] J. Hayfron-Acquah, M. S. Nixon, and J. Carter, “Automatic Gait Recognition by Symmetry Analysis,” Proc. Int’l Conf. Audio-and-Video-Based Biometric Person Authentication, pp. 272-277, 2001.
[6] A. Kale, N. Cuntoor, B. Yegnanarayana, A. N. Rajagopalan, and R. Chellappa, “Gait Analysis for Human Identification,” Proc. Int’l Conf. Audio-and-Video-Based Biometric Person Authentication, pp. 706-714, 2004.
[7] Y. Makihara, R. Sagawa, Y. Mukaigawa, T. Echigo, and Y. Yagi, “Gait Recognition Using a View Transformation Model in the Frequency Domain,” ECCV 2006.
[8] J. Han and B. Bhanu, “Individual Recognition Using Gait Energy Image,” IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 28, no. 2, pp. 316-322, Feb. 2006.


[9] S. Sarkar, P.J. Phillips, Z. Liu, I.R. Vega, P.Grother, and K.W. Bowyer, “The HumanID Gait Challenge Problem: Data Sets, Performance, and Analysis,” IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 27, no. 2, pp. 162-177, Feb. 2005.
[10] M. A. Hossain, Y. Makihara, W. Junqui, and Y, Yagi, “Clothes-Invariant Gait Identification Using Part-Based Adaptive Weight Control,” ICPR 2008.
[11] S. Singh and K.K. Biswas, “Spatio-temporal Energy based Gait Recognition,” IEEE International Conference on Data Mining, pp. 998-1003, 2009.
[12] Y. Pratheepan, J.V. Condell and G. Prasad, “The use of Dynamic and Static Characteristics of Gait for Individual Identification,” IEEE International Machine Vision and Image Processing, pp. 111-116, 2009.
[13] X. Yang, Y. Zhou, T. Zhang, G. Shu, and J. Yang, “Gait Recognition Based on Dynamic Region Analysis,” Signal Processing, vol. 88, no. 9, pp. 2350-2356, 2008.
[14] R. E. Bellman, “Adaptive Control Processes,” Princeton University Press, Princeton, NJ, 1961.
[15] The CASIA Gait Database http://www.cbsr.ia.ac.cn/
[16] The LIBSVM tools http://www.csie.ntu.edu.tw/~cjlin/libsvm/
[17] P. J. Phillips, H. Moon, S. Rizvi, and P. Rauss, “The FERET Evaluation Methodology for Face-Recognition Algorithms,” IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 22, no. 10, pp. 1090-1104, Oct. 2000.