臺灣博碩士論文加值系統

近年來隨著科技進步，臉部圖形辨識的應用越來越廣，在未來，臉部圖形辨識的使用及商業價值也越趨明顯，尤其在門禁系統、治安控管、或罪犯圖像建檔等。臉部圖形辨識有著特徵選取不易、辨識比較複雜及辨識耗費時間冗長等等狀況，許多需求及空間有待新方法來改善辨識精準度及辨識速度等最重要且最關鍵的問題。

本研究提出合理的特徵選取方法，即以改良式亂度和豪斯多夫距離為基礎，來解決特徵選取及比對的問題。特徵擷取採用灰階的不同來進行亂度計算，找出最合適亂度區間，將亂度效果作最好呈現。比對使用豪斯多夫距離，利用兩圖中每個點間的相互距離做幾何比對且此法特徵點過多或過少將導致計算複雜或錯誤辨識發生，故而，是一種最適合檢定本法是否能完成既定目標。

經過以五個樣本資料庫（每個樣本資料庫包含十種不同的呈現模式）中五十張皆為92*112的10kb樣本影像試驗，結果得出TypeⅠ error為8%左右及TypeⅡ error為15%左右的結果。而比對圖像須經過特徵選取及樣本比對兩個過程，使用MATLAB耗費時間總和約在1秒至1.5秒左右。

經由結果得知本實驗的方法，確實可再精進的發展、經得起更多試驗考證的方法，而且是一種可以藉合理亂度區間明顯降低計算量，使得擁有快速且相當精確度的臉部圖形辨識方法。

With the science and technology development in recent years, the application of face recognition is extensive. This technology will be useful and have commercial profit obviously in future, especially in entrance guard's system, the public security control, and establishing criminal's picture files. But face recognition has many problems such as difficult feature selection, complex recognition, and time-consuming so it requires new way which can improve the accuracy and speed to solve these important problems.

This research proposed a method which reasonably captures features to solve the problem by the improved entropy and Hausdorff distance. In according to counting entropy from different gray levels and finding the suitable entropy range, this method has presented a better result on feature selection. Comparing two images is to calculate Hausdorff distance which depends on counting the distance from each point in one image to all points in another. Because Hausdorff distance has very good reliability on the comparison of feature points, it will be the best way to see if the method in this research could reach the goal.

Through the experiment of 5 data samples with ten different facial expressions, and they are all 10k bytes sample images with resolution 92*112. The result showed that TypeⅠ error is about 8% and TypeⅡ error about 15%. Including feature capturing and sample recognition, the total time cost is about 1 to 1.5 seconds under MATLAB7.0.

Depending on this result, this face recognition method is a useful way to reduce the number of feature points and increase accuracy.

第一章、緒論

第一節 研究背景--------------------------------------------------------- 1
第二節 文獻回顧（圖形辨識）------------------------------------------- 2
第三節 文獻回顧（臉部辨識）------------------------------------------- 4
第四節 以亂度及豪斯多夫為基礎的辨識方法--------------------------- 7
第五節 論文結構------------------------------------------------ 10

第二章、方法

第一節 研究流程描述--------------------------------------------- 12
第二節 亂度（Entropy）----------------------------------------- 13
第三節 邊際取樣------------------------------------------------ 18
第四節 Hausdorff distance簡介---------------------------------- 18
第五節 實驗樣本------------------------------------------------ 21

第三章、結果 24


第四章、討論 33

參考文獻 36

[1]A.k. Jain, R.P. W. Duin and J.C. Mao. (2000). Statistical pattern recognition : A review. IEEE Transactions on Pattern Analysis and Machine Intelligence, 22（1）, 4-37.
[2]B. Srinivasa Reddy and B. N. Chatterji. (1996, AUGUST). An FFT-Based Technique for Translation, Rotation, and Scale-Invariant Image Registration. IEEE TRANSACTIONS ON IMAGE PROCESSING,5（8）.
[3]Barbara Zitova, Jan Flusser . (2003). Image registration methods: a survey. ELSEVIER Image and Vision Computing, 21, 977–1000,
[4]Canny, J.F.A. (1986). Computational approach to edge detection. IEEE Trans Pattern Analysis and Machine Intelligence, 8(6), 679-698.
[5] D. Y. Chen, H. Y. Mark Liao, H. R. Tyan, and C.W. Lin.(2005, November). Automatic Key Posture Selection for Human Behavior Analysis. IEEE International Workshop on Multimedia Signal Processing, Shanhai, China.
[6]Forrest M.Hoffman. An Introduction to Fourier Theory.
[7]H.L. Peng and S.Y. Chen.(1997).Trademark shape recognition using closed contours. Pattern Recognition, 18, 791-803.
[8] Huachun Tan, Yu-Jin Zhang.(2006). A novel weighted Hausdorff distance for face localization. ELSEVIER TImage and Vision Computing, 24, 656–662.
[9]J. B. Antoine Maintzand Max A. Viergever.（1998）. A Survey of Medical Image Registration. Medical Image Analysis, 2（1）, 1–37.
[10] J.H. Kim.(2005). Multiple Face Segmentation and Tracking Based on Robust Hausdorff Distance Matching. Journal of Applied Science, 740-744.
[11]K. Fukunaga.（1990）. Introduction to Statistical Pattern Recognition. Academic Press, Boston.
[12]Mira Park, Jesse S.J., Laurence S.W.(2003). Intelligent computer-aided diagnosis system for chest radiography. Elsevier Science B.V. and CARS, International Congress Series,1256, 1005– 1010.
[13] Oliver Jesorsky, Klaus J.K., and Robert W.F. (2001, June).Robust Face DetectionUsing the Hausdorff Distance. Third International Conference on Audio- and Video-based Biometric Person Authentication, Springer, Lecture Notes in Computer Science, LNCS-2091, 90–95.
[14] Peng Xiaoming,Chen Wufan.(2006). Hausdorff Distance Based Non-rigid Image Registration Using Feature Points. IEEE.
[15]R. Bajcsy and S. Kovacic.(1989).Multiresolution elastic matching. Computer Vision Graphics Image Processing, 46, 1-21.
[16]R. Jain, R. Kasturi and B. G. Schunck(1995), Machine Vision, McGraw-Hill Inc.
[17]Siwaphon Chunhavittayatera, Orachat Chitsobhuk, Kiatnarong Tongprasert.（2006 ,Feb）. Image Registration using Hough Transform and Phase Correlation. ICACT, 20-22,
[18]S. Haykin.（1999）. Neural Networks: A Comprehensive Foundation. Prentice-Hall, New Jersey.
[19] Shingo Iwano, Tatsuya Nakamura, Yuko Kamioka, Takeo Ishigaki.(2005). Computer-aided diagnosis: A shape classification of pulmonary nodules imaged by high-resolution CT . ELSEVIER Computerized Medical Imaging and Graphics, 29, 565–570.
[20]Shaojun Liu and Jia Li.(2006). Automatic medical image segmentation using gradient and intensity combined level set method. IEEE the 28th EMBS Annual International Conference.
[21] William J. Rucklidge.(1995). Locating Objects Using the Hausdorff Distance. IEEE computer society Proceedings of the Fifth International Conference on Computer Vision.
[22]W.S. Chen, S.W Shih, K.T. Shen, Lili Hsieh ,and K.H. Chr. Automatic Iris Recognition System based on Wavelet Transform and Energy Transform. Electrical Engineering, National Chi Nan University.
[23] Yann Labit, Johan Mazel.( 2008). HIDDEN: Hausdorff distance based Intrusion Detection approach DEdicated to Networks. IEEE computer society The Third International Conference on Internet Monitoring and Protection.
[24]Yu Tong, Hui Wang and Daoying Pi. ( 2006, June ). Fast Algorithm of Hough Transform-Based Approaches for Fingerprint Matching. IEEE , Proceedings of the 6th World Congress on Intelligent Control and Automation, 21 – 23.
[25] Face Recognition.( 2006, August ). National Science and Technology Council (NSTC),Committee on Technology, Committee on Homeland and National Security, Subcommittee on Biometrics,7.
[26]汪上曉（2004，5月）。資訊與熵。科學發展，337期。
[27]呂芳懌、余少棠、許閔雄、林晁立（ 1999，7月）。影像資料庫在人臉辨識上的應用— 前科犯資料庫。東海科學，1卷, A103-122頁 。台中市。
[28]宋淑蘭、洪添勝、王衛星、李震（ 2005，12月）。基於馬氏距離的荔枝圖像分割設計方法。瀋陽農業大學學報,36卷6期, A655-658頁。
[29]李棟良、梁振升、王於藩、陳家閔、曾鈺鈞、簡煒玲。人臉偵測與辨識系統。銘傳大學電腦與通訊工程學系，台北市。
[30]張紅穎、張加萬、孫濟洲（2006，十月）。基于混合互訊息的醫學圖像配准。計算機應用，第26卷10期
[31]彭子軒（2002）。慢顆粒流之輸送帶實驗與影像分析。碩士論文，國立中央大學土木工程研究所，桃園縣。
[32]彭國軒（2003）。快速物件辨認與定位－環狀樣板比對。碩士論文，清華大學。.
[33]鄧宜珍（2002）。遙測影像處理與地貌辨識。碩士論文，中央大學資訊工程研究所。
[34]龔威儒（2004）。局部遮蔽圖形之自動比對。碩士論文，清華大學工學院 動力機械工程學系。
[35]工業用影像辨識技術。工研院機械所。