臺灣博碩士論文加值系統

生物特徵分為生理特徵與行為特徵，生理特徵是與生俱來的，而行為特徵是從習慣動作中提取的，如字跡、步態、擊鍵。使用以上生物特徵作身份辨識稱為「生物辨識」，因應不同情況，使用的生物特徵也不同，如手機解鎖常使用指紋或臉部。現今身份辨識的技術大部分都是使用深度學習的方式，因其能夠減少光照、複雜背景對辨識結果的影響，讓身份辨識擁有更高的準確率與強健性(Robustness)。步態辨識是透過一個人走路的姿勢去做身份辨識，當臉部被遮蔽或不清晰則無法使用臉部辨識，步態辨識可以克服臉部辨識無法使用的狀況。本論文對步態辨識進行深入研究，提出一個基於人體建模特徵的步態辨識方法，使用人體姿勢估計模型從影片中輸出多幀人體骨架圖，再利用人體骨架圖做步態辨識。本方法與現今方法相比有更好的系統強健性(Robustness)，並且實作出一個可以對單一影片進行多人步態辨識的系統。

Biological characteristics are divided into physiological characteristics and behavioral characteristics. The former is owned by human being itself and the latter is extracted from habitual movement, such as hand writing, gait and keystroke. The use of biological characteristics for identification is called biometrics. Different biometrics are suitable for different situations. For examaple, fingerprints or faces are often used to unlock mobile phones. Currently, biometrics are based on deep learning methods as the recognition results will not be quite infected by light and complex background. Hence, biometrics have higher accuracy and robustness. Gait recognition means recognizing identity through walking posture. As we know, face recognition will fail while the face is obscured or blurred. Then gait recognition can overcome the unavailability of facial recognition. This thesis focuses on gait recognition and proposes a gait recognition method based on human body modeling features. The human body pose estimation model is used to output multiple frames of human skeleton images from the video, and then the human skeleton images are used for gait recognition. Compared to the current methods, our method has better robustness, and a gait recognition for multiple people on a single video is also proposed.

中文摘要 I
Abstract II
目錄 V
圖目錄 VII
表目錄 VIII
第一章 緒論 1
1.1 研究動機與目的 1
1.2 相關研究 1
1.2.1 步態辨識 1
1.2.2 人體姿勢估計 2
1.3 系統架構與相關硬體規格 3
1.3.1 系統架構 3
1.3.2 相關硬體規格 4
第二章 深度學習介紹與生物識別介紹 5
2.1 深度學習介紹 5
2.1.1 深度神經網路 5
2.1.2 卷積神經網路 5
2.2 生物識別介紹 6
2.2.1 生理特徵識別 7
2.2.2 行為特徵識別 8
第三章 步態辨識研究介紹 11
3.1 基於人體外觀特徵做辨識(Appearance-based) 11
3.2 基於人體建模做辨識(Model-based) 14
第四章 研究方法與成果展示 16
4.1 研究流程說明 16
4.2 網路介紹與資料集介紹 17
4.2.1 人體姿勢估計模型 17
4.2.2 步態辨識模型 18
4.2.3 資料集介紹 22
4.3 研究方法 22
4.3.1 步態辨識 23
4.3.2 多人步態辨識 27
4.4成果展示 29
第五章 結論與未來展望 35
參考文獻 36

[1]Z. Cao, G. Hidalgo, T. Simon, S.-E. Wei, and Y. Sheikh, "OpenPose: Realtime multi-person 2D pose estimation using part affinity fields," IEEE transactions on pattern analysis machine intelligence, vol. 43, no. 1, 2019, pp. 172-186.
[2]H.-S. Fang, S. Xie, Y.-W. Tai, and C. Lu, "Rmpe: Regional multi-person pose estimation," IEEE International Conference on Computer Vision, 2017, pp. 2334-2343.
[3]K. Sun, B. Xiao, D. Liu, and J. Wang, "Deep high-resolution representation learning for human pose estimation," IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019, pp. 5693-5703.
[4]L. Wang, T. Tan, W. Hu, and H. Ning, "Automatic gait recognition based on statistical shape analysis," IEEE transactions on image processing, vol. 12, no. 9, 2003, pp. 1120-1131.
[5]J. Han and B. Bhanu, "Individual recognition using gait energy image," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 28, no. 2, 2005, pp. 316-322.
[6]H. Chao, Y. He, J. Zhang, and J. Feng, "Gaitset: Regarding gait as a set for cross-view gait recognition," AAAI conference on artificial intelligence, vol. 33, no. 01, 2019, pp. 8126-8133.
[7]C. Fan et al., "Gaitpart: Temporal part-based model for gait recognition," IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020, pp. 14225-14233.
[8]S. Yu, D. Tan, and T. Tan, "A framework for evaluating the effect of view angle, clothing and carrying condition on gait recognition," in 18th International Conference on Pattern Recognition (ICPR'06), vol. 4, 2006, pp. 441-444.
[9]R. He, T. Tan, and L. Wang, "Robust recovery of corrupted low-rankmatrix by implicit regularizers," IEEE transactions on pattern analysis machine intelligence, vol. 36, no. 4, 2013, pp. 770-783.
[10]S. Zheng, J. Zhang, K. Huang, R. He, and T. Tan, "Robust view transformation model for gait recognition," in 2011 18th IEEE International Conference on Image Processing, 2011, pp. 2073-2076.
[11]W. An et al., "Performance evaluation of model-based gait on multi-view very large population database with pose sequences," IEEE Transactions on Biometrics, vol. 2, no. 4, 2020, pp. 421-430.
[12]R. Liao, S. Yu, W. An, and Y. Huang, "A model-based gait recognition method with body pose and human prior knowledge," Pattern Recognition, vol. 98, 2020, p. 107069.
[13]S. Hochreiter, Y. Bengio, P. Frasconi, and J. Schmidhuber, "Gradient flow in recurrent nets: the difficulty of learning long-term dependencies," ed: A field guide to dynamical recurrent neural networks. IEEE Press, 2001.
[14]Y. LeCun, L. Bottou, Y. Bengio, and P. Haffner, "Gradient-based learning applied to document recognition," in Proceedings of the IEEE, vol. 86, no. 11, 1998, pp. 2278-2324.
[15]A. Krizhevsky, I. Sutskever, and G. E. Hinton, "Imagenet classification with deep convolutional neural networks," Advances in neural information processing systems, vol. 25, 2012, pp. 1097-1105.
[16]A. K. Jain, R. Bolle, and S. Pankanti, "Biometrics: Personal identification in networked society," Springer Science & Business Media, 2006.
[17]Y. Taigman, M. Yang, M. A. Ranzato, and L. Wolf, "Deepface: Closing the gap to human-level performance in face verification," IEEE conference on computer vision and pattern recognition, 2014, pp. 1701-1708.
[18]T. Ahonen, A. Hadid, and M. Pietikainen, "Face description with local binary patterns: Application to face recognition," IEEE transactions on pattern analysis and machine intelligence, vol. 28, no. 12, 2006, pp. 2037-2041.
[19]E. Variani, X. Lei, E. McDermott, I. L. Moreno, and J. Gonzalez-Dominguez, "Deep neural networks for small footprint text-dependent speaker verification," in 2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2014, pp. 4052-4056.
[20]Y. Fu et al., "Horizontal pyramid matching for person re-identification," AAAI Conference on Artificial Intelligence, vol. 33, no. 01, 2019, pp. 8295-8302.
[21]F. Schroff, D. Kalenichenko, and J. Philbin, "Facenet: A unified embedding for face recognition and clustering," IEEE conference on computer vision and pattern recognition, 2015, pp. 815-823.
[22]C.-H. Chen and D. Ramanan, "3d human pose estimation= 2d pose estimation+ matching," IEEE Conference on Computer Vision and Pattern Recognition, 2017, pp. 7035-7043.
[23]C. Ionescu, D. Papava, V. Olaru, and C. Sminchisescu, "Human3. 6m: Large scale datasets and predictive methods for 3d human sensing in natural environments," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 36, no. 7, 2013, pp. 1325-1339.
[24]C. Chen, J. Liang, H. Zhao, H. Hu, and J. Tian, "Frame difference energy image for gait recognition with incomplete silhouettes," Pattern Recognition Letters, vol. 30, no. 11, 2009, pp. 977-984.
[25]J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, "You only look once: Unified, real-time object detection," IEEE conference on computer vision and pattern recognition, 2016, pp. 779-788.
[26]L. Wang, T. Tan, H. Ning, and W. Hu, "Silhouette analysis-based gait recognition for human identification," IEEE transactions on pattern analysis machine intelligence, vol. 25, no. 12 2003, pp. 1505-1518,.
[27]D. Tan, K. Huang, S. Yu, and T. Tan, "Efficient night gait recognition based on template matching," in 18th International Conference on Pattern Recognition (ICPR'06), vol. 3, 2006, pp. 1000-1003.
[28]A. Bewley, Z. Ge, L. Ott, F. Ramos, and B. Upcroft, "Simple online and realtime tracking," in 2016 IEEE international conference on image processing (ICIP), 2016, pp. 3464-3468.
[29]T. Teepe, A. Khan, J. Gilg, F. Herzog, S. Hörmann, and G. Rigoll, "GaitGraph: Graph convolutional network for skeleton-based gait recognition," arXiv preprint arXiv:.11228, 2021.
[30]K. He, G. Gkioxari, P. Dollár, and R. Girshick, "Mask r-cnn," IEEE international conference on computer vision, 2017, pp. 2961-2969.