臺灣博碩士論文加值系統

人臉辨識系統的好壞取決於特徵之擷取，而樣本比對(template matching)與幾何特徵(geometric feature)為兩種主要的方法。本論文將著重於樣本比對的方法上。而一般2D人臉辨識(face recognition)，其主要的變化參數為照度(illumination) 、臉的位置(face position)與觀測方向(viewing direction)，照度於二維影像則是灰階值強度。本文以光學立體法 (Photometric Stereo Method, PSM) 重建出3D的立體人臉做為人臉辨識的資料庫，與2D人臉的差異在於變化參數為人臉的深度資訊 (depth information)，當然觀測方向與臉的位置亦是立體人臉的變化參數之一。傳統上使用的主要成份分析法 (Principal Components Analysis, PCA) 應用在人臉辨識領域中已在電腦視覺(computer vision)、工業機器人(industrial robotics)等方面皆有不錯的結果。文中我將提出以結合小波轉換與主要成份分析法相較於小波轉換 、主要成份分析法所擷取之人臉特徵，分析比較其在立體人臉與2D人臉辨識的效果，並結合最近中心 (Nearest Center, NC)、最近特徵線 (Nearest Feature Line, NFL) 與線性鑑別分析 (Linear Discriminant Analysis, LDA) 等分類器或決策法則，以加強辨識系統的強健性，我的結果將會顯示出小波轉換 (wavelet transform )在人臉特徵擷取上的優越效能。

The methods of feature extraction are most importance in a face recognition system. There are two representation methods, template matching and geometric feature. In the 2D face recognition, the recognition rate is depended on the illumination, face location and viewing direction. The gray-value in the 2D face image is due to the illuminated intensity. This thesis studies the face recognition using the 3D faces which are reconstructed by Photometric Stereo Method. Our images contain with the depth-information which reconstructed in 3D faces. They are different to 2D gray-value images. We know that Principal Component Analysis method has better performance in face recognition. And it had been applied in computer vision, industrial robotics. In this thesis, we will present the novel approach which combines wavelet and PCA to generate the face feature. Furthermore, we will employ this face feature to compare with both in wavelet space and PCA space for face recognition, and compare the results. Our results will show the comparison results of different classifiers, for example, Nearest Center, Nearest Feature Line and Linear Discriminant Analysis, we will show that this new approach reveals the more excellent performance in face recognition.

摘要 i
ABSTRACT iii
目錄 v
圖目錄 vii
表目錄 viii
第1章 緒 論 1
1.0 研究動機與目的 1
1.1 系統流程與架構 1
1.2 本文架構 2
第2章 人臉特徵擷取 5
2.0 前言 5
2.1 PSM方法之立體人臉影像重建 5
2.1.1 光學式立體法 (PSM) 之演算法 6
2.1.2 PSM重建方法之限制條件 6
2.1.3 立體人臉模型與深度影像 7
2.2 主要成份分析參數擷取法 8
2.3 幾何特徵參數擷取法 9
2.4 小波轉換與結合主要成份分析之特徵參數擷取法 11
2.4.1 小波轉換特徵擷取法 11
2.4.2 小波轉換結合PCA特徵擷取法 11
第3章 小波轉換 13
3.0 前言 13
3.1 數學基礎 13
3.2 多層次解分析 15
3.3 小波轉換 16
3.4 二維小波轉換 18
第4章 特徵參數分類與決策法則之應用 20
4.0 前言 20
4.1 最近中心決策法(Nearest Center) 20
4.2 最近特徵線決策法(Nearest Feature Line Method) 21
4.3 自組式映射類神經網路 22
4.4 線性鑑別分析法 24
第5章 程式模擬結果 26
5.0 前言 26
5.1 人臉資料庫 26
5.2 實驗架構 30
5.3 模擬結果比較 31
5.3.1 使用主要成份分析法 31
5.3.2 使用小波轉換 33
5.3.3 結合小波轉換與主要成份分析法 34
5.4 以3D資料庫做識別族群外人臉之模擬實驗 37
第6章 討論與未來研究方向 39
6.0 討論 39
6.1 未來研究方向 40
參考文獻 41

[1] Rafael C. Gonzalez, Richard E. Woods, “Digital Image Processing”, Addison Wesley, 1993.
[2] Matthew A. Turk and Alex P. Pentland, “Face Recognition Using Eigenfaces”, in Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition, Hawaii, June 1991, pp. 586--591.
[3] 沈良澤, “人臉辨識之研究”網站, http://www.csie.nctu.edu.tw/~ljshen/ccl/ccl.html
[4] Stan Z. Li and Juwei Lu, “Face Recognition Using the Nearest Feature Line Method”, IEEE Transactions on Neural Networks, vol.10, no2,March 1999.
[5] Guodong Guo, Stan Z. Li, and Kap Luk Chan, “Face Recognition by Support Vector Machines”, Image and Vision Computing, vol. 19, Issue 9-10, 631-638, 2001.
[6] Alberto Tomita, Jr. and Rokuya Ishii, “Determining the Orietation of a Person’s Hand by Using the Photometric Stereo Method”, Proceedings of IECON’96, Volume: 1, 5-10 Aug 1996.
[7] Erik Hjelmas, “Face Detection: A Aurvey”, Computer Vision and Understanding 83, 236 — 274, 2001.
[8] Anders Eriksson, David Weber, “Towards 3-dimensional Face Recognition” IEEE Proceedings of 5th ICON, pp.401-406, 1999.
[9] Gunter Bellaire and Karsten Schluns, 3-D Object Recognition by Matching the Total View Information”, IEEE, Proceedings of ICPR’96, Vinna, Astria, Aug. 25-29:534-539, 1996.
[10] Yu Hengyong, Wang Yong, Mou Xuanqin, Cai Yuanlong, “Face Detection Model Based on Distance Measure of Regional Feature”, IEEE Proceedings of 5th International Conference on Signal Processing, vol.3, pp.1479-1482, 2000.
[11] Roberto Brunelli and Tomaso Poggio, “Face Recognition: Feature versus Templates”, IEEE Transactions on Pattern Analysis and Machine Intelligence. Vol. 15, no. 10, oct 1993.
[12] C.D. Caleanu , “Facial Recognition Using Committee Neural Network”, IEEE Proceedings of the 5th Seminar on Neural Network Applications in Electrical Engineering, Page(s): 97 -100, Set. 25-27, 2000.
[13] Jaideva C. Goswami, Andew K. Chan, “Fundamentals of Wavelets — Theory, Algorithms and Applications.” Wiley 1999.
[14] Graps, A. “An Introduction to Wavelets”, IEEE Computational Science and Engineering, Volume: 2 Issue: 2, summer 1995, Page(s): 50 -61.
[15] C. Sidney Burrus, Ramesh A. Gopinath, and Haitao Guo, “Introduction to Wavelets and Wavelet Transforms”, Prentice-Hall International.
[16] Aleix M. Martinez, Avinash C. Kak, “PCA Versus LDA”, IEEE Transaction on Pattern Analysis and Machine Intelligence. vol.23 no.2 Feb 2001.
[17] Daniel L. Swets and John Weng, “Using Discriminant Eigenfeatures for Image Retrieval”, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol.18 no.8, August 1996.
[18] Andrew Webb, “Statistical Pattern Recognition”, London : Arnold , 1999.
[19] Reinhard Klette, Karsten Schluns, Andreas Koschan, “Computer Vision — Three-Dimensional Data from Images”, Springer 1996.
[20] Ernest L. Hall, ”Computer Image Processing and Recognition”.
[21] Earl Gose, Richard Johnsonbaugh, Steve Jost, “Pattern Recognition and Image Analysis”, Prentice Hall.
[22] Chia-Yen Chen, Reinhard Klette and Ramakrishna Kakarala, “Albedo Recovery Using a Photometric Stereo Approach”, Pattern Recognition, IEEE Proceedings of 16th Conference on Pattern Recognition, vol.1,pp.700-703, 2002,.
[23] Q. Wu, Z. Xiong, Y. Wang, T. Chen, and K. R. Castleman, “On Optimal Subspace for Appearance-base Object Recognition”, Proceedings of ICIP’02, Rochester, New York, Sep.22-25, 2002.
[24] Yu-shan Tseng, Chi-Fa Chen,Chia-Yen Chen. "Face Recognition Based on Principal Component Analysis of 3D Images" Image and Video Computing New Zealand 2002.
[25] 曾裕山,陳基發,李建興, "以小波轉換做3D人臉辨識”, 自動控制與電力研討會，session E1, pp.16-21, 2002.
[26] Andrew Bruce, Hong-Ye Gao, “Applied Wavelet Analysis with S-PLUS”, Springer 1996.
[27] 曾裕山,陳基發, ”結合PCA與小波轉換之人臉辨識”, 電機資訊工程研討會，page(s):60-63, 2002。