在身分識別的領域中，辨識效能常受限於資料品質與環境的影響。理想的做法是事先考慮到這些因素，並逐一解決，但要完善地蒐集到這些資訊幾乎是不可能的。在本論文，我們提出一種漸進式學習模型，應用於基於影像特徵的多角度人物識別系統，在訓練階段只需準備少量關於受測者的特徵訓練基礎分類器，在辨識階段，被擷取到的特徵除了進行身分識別，系統能自動根據基礎知識從這些特徵裡選取有用的進行學習，如此辨識率能隨著知識的豐富而逐漸提升。被擷取到的特徵裡，可能存在錯誤的，或是過於相似的特徵，這些特徵被系統學習後反而會導致辨識率下降以及增加系統負擔，所以在學習模型裡，我們基於辨識結果的可信度和集中度設計特徵選擇策略與身分確認機制，以解決此類問題。另外，從不同角度觀察人的外表其差異非常大，故對觀察角度的變化必須獨立處理。在進行身分識別之前，以輪廓、膚色區域與外型特徵結合模板比對技術，加入身體方向估測的前處理，以達到多角度之人物識別。在實驗部分，首先評估身體方向估測方法的可信度，平均估測正確率達到98.44%以上。接著對正面、左側面、右側面三種方向分別比較有無採用漸進式學習機制之身分識別結果，最後結合這些方向的辨識結果，實驗數據說明學習機制能提升13.09%的平均辨識率。

In the field of person identification, the performance is almost limited by the quality of data and influence of environment. Besides, it is hard to consider and overcome all variations of data in advance. In this thesis, an evolutionary learning model which can automatically utilize basic information to absorb related information is proposed and applied in multi-view person identification. The performance can be improved because knowledge gets richer. In order to avoid wrong information being learnt, a tutor scheme is necessary. Therefore, we propose Feature Selection Strategy (FSS) and identity ensuring mechanism. The former can sieve out the redundant and wrong information, and the later ensures the recognition result to avoid wrong learning. They are driven by reliability and recognition concentration estimation. Because the appearance is very diverse in different body directions, a person orientation estimation method is necessary to identify the orientations of subjects first. Template matching is adopted to estimate the person orientations. In addition to contours, shapes and skin color areas provide more orientation information for templates. In the experiments, the performance of person orientation estimation is shown, and average accuracies are higher than 98.44%. Then, the accuracy of person identification in three orientations, front, left profile and right profile, with and without evolutionary learning model are evaluated separately. At last, the identifications in different orientations are merged. The result illustrates the evolutionary learning helps to increase average accuracy about 13.09%.

[1] J. Hirata, M. Morimoto, and K. Fujii, “Estimating Face Direction from Low Resolution Images,” World Automation Congress, 2008.
[2] X. B. Cao, H. Qiao, and J. Keane, “A Low-Cost Pedestrian-Detection System with a Single Optical Camera,” IEEE Transactions on Intelligent Transportation Systems, vol. 9. No. 1, March 2008, pp.58-67.
[3] E. Borenstein, E. Sharon and S. Ullman, “Combining Top-down and Bottom-up Segmentation,” IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshop, 2004.
[4] J. Okubo, B. Sugandi, H. Kim, J. K. Tan, and S. Ishikawa “Face Direction Estimation Based on Eigenspace Technique,” International Conference on Control, Automation and Systems, Korea, 2008.
[5] X. P. Chen, Q. Q. Yang, H. H. Liao, W. P. Sun, and S. S. Yu, “Real-Time Face Pose Estimation in Video Sequence,” 2nd International Workshop on Education Technology and Computer Science, 2010.
[6] M. W. Lee, and R. Nevatia, “Body Part Detection for Human Pose Estimation and Tracking,” IEEE Workshop on Motion and Video Computing, 2007.
[7] A. Genco, and E. Varrica, “Human Figure Recognition and the Estimation of its Direction,” 7th International Workshop on Computer Architecture for Machine Perception, 2005.
[8] A. Mclvor, “Background Subtraction Techniques,” in Proc. of Image Video Computing, 2000, pp. 147-153.
[9] L. J. Li, and F. F. Li, “OPTIMOL: Automatic Online Picture Collection via Incremental Model Learning,” International Journal of Computer Vision, Vol. 88, No. 2, 2010, pp. 147-168.
[10] M. F. Balcan, A. Blum, P. P. Choi, J. Lafferty, B. Pantano, M. R. Rwebangira, and X. Zhu, “Person Identification in Webcam Images: An Application of Semi-Supervised Learning,” Proc. Of the 22th ICML Workshop on Learning with Partially Classified Training Data, Bonn, Germany, 2005.
[11] T. Kidera, and S. Ozawa, “An Incremental Learning Algorithm of Ensemble Classifier Systems,” International Joint Conference on Neural Networks, July 2006.
[12] S. Ozawa, S.L. Toh, S. Abe, S. Pang, and N Kasabov, “Incremental Learning of Feature Space and Classifier for Face Recognition,” Neural Network, Volume 18, Issue 5-6, July-August 2005, pp. 575-584.
[13] A. Almaksour and E. Anquetil, “Fast Incremental Learning Strategy Driven by Confusion Reject for Online Handwriting Recognition,” 10th International Conference on Document Analysis and Recognition, 2009.
[14] R. Zhang, and A.I. Rudnicky “A New Data Selection Principle for Semi-Supervised Incremental Learning,” 18th ICPR, 2006.
[15] R. Cappelli, D. Maio, and D. Maltoni, “Subspace Adaption for Incremental Face Learning,” 7th Internal Conference on Control, Automation, Robotics and Vision, Singapore, 2002.
[16] “OpenCV,” Software available at http://sourceforge.net/projects/opencvlibrary/
[17] K. Ohba, Y. Sato, and K. Ikeuchi, “Appearance-based Visual Learning and Object Recognition with Illumination Invariance,” Machine Vision and Applications, 2000, pp. 189-196.
[18] A. Yilmaz, O. Javed, and M. Shah, “Object tracking: A survey,” ACM Computing Surveys, Vol. 38, No. 4, 2006, pp. 1-45.
[19] J. Shi and C. Tomasi, “Good Features to Track,” in Proc. IEEE Computer Society Conf. on Computer Vision and Pattern Recognition, 1994, pp. 593-600.
[20] B.D. Lucas, and T. Kanade, “An Iterative Image Recognition Technique with an Application to Stereo Vision,” Proceedings of Imaging Understanding Workshop, 1981, pp. 121-130.
[21] C. Garcia, and G. Tziritas, “Face Detection Using Quantized Skin Color Regions Merging and Wavelet Packet Analysis,” IEEE Transactions on Multimedia, Vol. 1, No. 3. September 1999.
[22] R. Lukac, and K. N. Plataniotis, “Color Image Processing Methods and Applications,” 2007, pp. 294-307.
[23] S. K. Singh, D. S. Chauhan, M. Vatsa, and R. Singh, “A Robust Skin Color Based Face Detection Algorithm,” Tamkang Journal of Science and Engineering, Vol. 6, No. 4, 2003, pp. 227-234.
[24] P. Viola, and M. Jones, “Robust Real-time Object Detection,” 2nd International Workshop on Statistical and Computational Theories of Vision – Modeling, Learning, Computing, and Sampling, Vancouver, Canada, July 13, 2001.
[25] L.F. Chen, H.Y.M. Liao, M.T. Ko, J.C. Lin, and G.J. Yu, “A new LDA-based face recognition system which can solve the small sample size problem,” Pattern Recognition, Vol. 33, No. 10, 2000, pp. 1713-1726.
[26] M. Turk, and A. Pentland, “Face Recognition using Eigenfaces,” in Proceedings of the Computer Society Conference on Computer Vision and Pattern Recognition, 1991, pp. 586-591.
[27] C. BenAbdelkader, R. Cilter, and L. Davis, “Person Identification using Automatic height and stride estimation,” in Proc. Int. Conf. Pattern Recognition, Vol. 4, Canada, 2002, pp. 377-380.
[28] Y. Iwashita and R. Kurazume, “Person Identification from Human Walking Sequences using Affine Moment Invariants,” Proc. IEEE Int. Conf. Robotics and Automation, 2009, pp. 436-441.
[29] W. Zhao, R. Chellappa, A. Rosenfeld, and P. J. Phillips, “Face Recognition: A Literature Survey,” ACM Computing Surveys, Vol. 35, No. 4, December 2003, pp. 399-458.
[30] W. Huang, B.H. Lee, L. Li, and K. Leman, “Face Recognition by Incremental Learning,” in Proc. IEEE Int. Conf. System, Man and Cybernetics, Vol. 5, 2003, pp. 4718-4723.
[31] L. Huang, J. Yao, and Y. P. Tan, “Enhancing Incremental Learning/Recognition via Efficient Neighborhood Estimation,” in ICME, 2010, pp. 637-642.
[32] W. S. Kang, J. H. Na, H. S. Ahn, and J. Y. Choi, “Face Recognition System by Fast and Incremental Learning Method,” in SICE-ICASE, International Joint Conference, Busan, 2006.
[33] P. N. Belhumeur, J. P. Hespanha, and D. J. Kriegman, “Eigenfaces vs. Fisherfaces: Recognition Using Class Specific Linear Projection,” IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 19, no 7, July 1997, pp. 711-720.