本論文的研究動機在於如何將圖樣辨識理論應用於海洋科技上，並探討海洋科技之兩個研究主題，分別為水上雷達目標辨識以及水下通訊定位。
本研究的第一部份（詳見第三章）主要探討「利用圖樣辨識理論來實現水上雷達目標辨識」，在本論文的水上雷達目標辨識研究中，是以船艦為例。我們利用船艦目標的電磁波雷達散射截面積（Radar Cross Section, RCS）來當作辨識依據，而電磁波RCS本身是一個包含時間、頻率、方位角和極化方向的函數，這些資料通常過於龐大且無規則性，不易於儲存分析辨識。因此，本論文為了簡化RCS散射資料以利於雷達目標辨識，我們將圖樣辨識理論中的主成份分析（Principal Component Analysis, PCA）和線性鑑別式分析（Linear Discriminant Analysis, LDA）應用到水上雷達目標辨識中，我們擷取了船艦的RCS散射資料的特徵來做辨識。本論文的RCS是以商用電磁軟體Ansoft HFSS模擬而得，而我們也考慮到在現實環境中雜訊對水上雷達目標辨識的影響，所以我們也主動將模擬而得的RCS加入隨機雜訊，以更貼近實際量測資料。雜訊的存在，非常不利於雷達目標辨識，所以我們利用奇異值分解（Singular Value Decomposition, SVD）搭配Hankel法來分離出雷達訊號中的雜訊。最後為了使本論文的水上雷達目標辨識更加完善，我們利用分類與迴歸樹（Classification And Regression Tree, CART）搭配PCA來提高水上雷達目標辨識率。由最後模擬的結果可以得知本論文所提出的水上雷達目標辨識能夠達到高辨識率，驗證了本論文將圖樣辨識理論應用在此研究主題的可行性。
本研究的第二部份（詳見第四章）主要探討「利用圖樣辨識理論來實現水下通訊定位」。本論文所探討的水下通訊定位，不同於一般水下技術如長基線、短基線、…之類的商用儀器所稱之水下定位。我們的定位是要將定位流程與結果完全融入通訊系統中，讓通訊系統隨時可以透過通訊的訊號來即時計算出Client端（即無線通訊的使用者）的位置，讓Client端的位置時時刻刻被通訊系統所掌握，這種型式的水下通訊定位，目的是在為將來的水下無線網路之多基地台通訊的context-aware service作準備。水下通訊定位其實就如同陸地行動通訊網路中，手機使用者的位置隨時會被基地台的通訊管理系統所掌握一樣，通訊系統如能掌握Client端使用者的位置，當Client端使用者移動跨越基地台管轄區時，基地台作換手交接（hand-off）時就不會延遲通訊。在本論文的水下通訊定位中，我們直接從Client端使用者的水下通訊接收機（本論文以水下麥克風來模擬）和其接收的水下通訊訊號（本論文以魚探機所發射的AM振幅調變訊號來模擬）計算出Client使用者的位置。這樣的通訊定位能讓通訊Client端的使用者不須再隨身攜帶額外的定位設備就能夠定位，並將定位結果即時融入通訊系統。現今傳統的水下定位儀器大多必須量測訊號的到達時間差或方向，若訊號遇到多重路徑反射，這些反射訊號會很難被處理且會影響到定位的準確性。為了克服這個難題，本論文在定位的過程中引用指紋特徵比對法的概念，應用此概念的優點在於我們能夠免除反射訊號對定位結果所帶來的負面影響。此概念的定位流程就像是比對人類指紋一樣，先分為兩個階段，分別是收集資料的離線狀態和實際定位的線上狀態。在離線狀態中，我們在不同取樣位置接收水下通訊的訊號並儲存至資料庫。在線上狀態中，我們將接收到的即時訊號與資料庫的訊號做比對，利用機率統計概念的最大似然法（Maximum Likelihood）來估算出目前接收訊號者的所在位置。接著我們再利用主成份分析PCA或線性鑑別式分析LDA對水下接收訊號做前置處理，將水下接收訊號投影到特徵空間來降低水下接收訊號的維度，如此一來便能夠減少定位所需的計算量，並可藉由投影來增加量測訊號的可辨識度。然而在實際水下環境中，訊號常會夾帶著雜訊，為了消除水下接收訊號的雜訊，我們利用奇異值分解SVD及Hankel法來分離出水下接收訊號的雜訊。最後在定位演算法中，我們加入了分類與迴歸樹CART演算法，CART的運算速度快，能夠節省在實際定位中的所需計算時間，而且也能夠達到維持很高的定位準確率。
本論文共分五章。第一章為緒論，介紹研究背景、動機、文獻回顧、論文架構及研究貢獻。第二章介紹圖樣辨識理論介紹，包括主成份分析（PCA）、線性鑑別式分析（LDA）、奇異值分解（SVD）及Hankel法、分類與迴歸樹（CART）。第三章是本研究的第一部份，主要研究「利用圖樣辨識理論來實現水上雷達目標辨識」。第四章是本研究的第二部份，主要研究「利用圖樣辨識理論來實現水下通訊定位」。第五章是結論。

In this dissertation, pattern recognition techniques are applied to two topics of oceanic technologies, including air radar target recognition and underwater communication localization.
In the first part of this study, pattern recognition techniques are applied to air radar target recognition. We utilize models of ships as targets for identification. The data of radar cross section (RCS) scattered from different types of targets are collected for identification. In general, these RCS data are very complex and are thus difficult to implement recognition. To enhance the recognition ability, we project the RCS data into feature space of PCA (principal components analysis) or LDA (linear discriminant algorithm). The RCS data in this study are obtained through the commercial tool of Ansoft-HFSS software. The original simulated data contain no noise. To make the simulated RCS more consistent with the practical experiment, random noises are added to the simulated data. The existence of random noise will degrade the performance of radar target recognition. In this study, the SVD (singular value decomposition) with Hankel method is applied to separate the noise part from the clean part of RCS data. In addition, the method of CART (classification and regression tree) is combined with the PCA method to give alternative approach of radar target recognition.
In the second part of this study, pattern recognition techniques are applied to underwater communication localization. The communication localization means that the localization is included into the communication system. In other words, the location of the client terminal will be obtained in the communication system by communication signals. Conventional underwater localization systems often utilize approaches based on time of arrival (TOA) or direction of arrival (DOA). Unfortunately, the performances of such approaches are affected by the multi-path reflection. In addition, such approaches require expensive hardware to obtain accurate measurement. To overcome these drawbacks, our localization is based on probabilistic fingerprinting approaches. The approaches are divided into two stages, i.e., the off-line (training) and on-line (predicting) stages. In the off-line stage, signals collected by the sound receiver at different sampling locations are stored to constitute the database (i.e., signal map). In the on-line stage, the real-time signal is measured and the location is thus estimated by comparing the real-time signal with the database. Our localization scheme is based on the probabilistic pattern recognition of acoustic communication signals, but not on ray tracing of signal propagation. Therefore, our underwater localization scheme is not affected by reflected signals. It can tolerate multi-path signals. This will greatly reduce both the hard-ware cost and calibration difficulty in measurement. To enhance the recognition ability of location estimation and reduce the data complexity, all received signals are projected onto the feature space of PCA and LDA. Each projected feature is assumed to have Gaussian probabilistic distributions. Therefore, the location information can be easily obtained by pattern recognition of projected features in PCA and LDA space. In practical measurement, there are severe noisy effects due to the uncertain characteristics in underwater environments. To reduce such noisy effects, the measured signals are processed by the SVD with Hankel technique. Finally, we also utilize the CART technique to obtain efficient pattern matching in the on-line stage. Experiments were successfully conducted in a bounded water pool to verify the benefits of our underwater localization schemes.
This dissertation is divided into five chapters. Chapter 1 gives the introduction of this study. Chapter 2 gives theoretical formulations of pattern recognition utilized in this study. The first part of this study, i.e., the application of pattern recognition to air radar target recognition, is given in Chapter 3. The second part of this study, i.e., the application of pattern recognition to underwater communication recognition, is given in Chapter 4. Finally, the conclusion is given in Chapter 5.

[1] Hajduch G., Le Caillec JM, and Garello R, “Airborne high-resolution ISAR imaging of ship targets at sea”, IEEE Transactions on Aerospace and Electronic Systems, Vol. 40, No. 1, pp. 378-384, 2004.
[2] Musman S, Kerr D, and Bachmann C, “Automatic Recognition of ISAR Ship Image”, IEEE Trans. Aerospace Electronic Systems, Vol. 32, Issue 4, pp. 1392-1403, 1996.
[3] Menon M, “Automatic Ship Classification System for ISAR Imagery”, SPIE Proc. Applications and Science of Artificial Neural Networks, Vol. 2492, pp. 373-388, 1995.
[4] Osman H, Blostein S, and Gagnon L, “Classification of Ships in Airborne SAR Imagery Using Backpropagation Neural Networks”, SPIE Proc. Radar Processing, Technology, and Applications II , Vol. 3161, pp. 126-136, 1997.
[5] Tello M, Lopez-Martinez C, and Mallorqui JJ, “A novel algorithm for ship detection in SAR imagery based on the wavelet transform”, IEEE Geoscience and Remote Sensing Letters, Vol. 2, No. 2, pp. 201–205, 2005.
[6] Chen Y, Yin J, Chai X, and Yang Q, “Power-efficient access point selection for indoor location estimation”, IEEE Transactions on Knowledge and Data Engineering, Vol. 18, No. 7, pp. 887-888, 2006.
[7] Kaemarungsi K and Krishnamurthy P, “Properties of indoor received signal strength for WLAN location fingerprinting”, Proceedings of the first annual international conference on mobile and ubiquitous systems: networking and service, 2004.
[8] Kaemarungsi K and Krishnamurthy P, “Modeling of indoor positioning systems based on location fingerprinting”, Proceedings of the IEEE INFOCOM, 2004.
[9] Lin TN and Lin PC, “Performance comparison of indoor positioning techniques based on location Fingerprinting in wireless networks”, Wireless Networks, Communications and Mobile Computing, pp. 1569-1574, 2005.
[10] Roots T, Myllymaki P, and Tirri H, ”A statistical modeling approach to location estimation”, IEEE Transactions on Mobile Computing, Vol. 1, Issue 1, pp. 59-69, 2002.
[11] Taheri A, Singh A, and Agu E, “Location fingerprinting on infrastructure 802.11 wireless local area networks (WLANs) using locus”, Proceedings of the 29th annual IEEE international conference on local computer networks, 2004.
[12] Youssef MA, Agrawala A, and Shankar AU, “WLAN location determination via clustering and probability distributions”, Pervasive Computing and Communications, 2003.
[13] Fang SH, Lin TN, and Lin PC, “Location Fingerprinting in a Decorrelated Space”, IEEE Transactions on Knowledge and Data Engineering, Vol. 20, Issue 5, pp. 685-691, 2008.
[14] Novak LM and Owirka GJ, “Radar Target Identification Using an Eigen-Image Approach”, Proc. of the IEEE National Radar Conference, pp. 129-131, 1994.
[15] Sirovich L and Kirby M, “Low-Dimensional Procedure for the Characterization of Human Faces”, Journal of the Optical Society of America A: Optics, Image Science, and Vision, Vol. 4, pp. 519-524, 1987.
[16] Singstock BD, Rogers SK, and Kabrisky M, “Automatic Target Recognition Using Karhunen-Loève Transform Generated Eigenimages”, SPIE Proc. Signal Processing, Sensor Fusion, and Target Recognition, Vol. 1699, pp. 232-240, 1992.
[17] Turk M and Pentland A, “Eigenfaces for Recognition”, Journal of Cognitive Neuroscience, Vol. 3, pp. 71-86, 1991.
[18] Theodoridis S and Koutroumbas K, Pattern Recognition, 2nd edition, Academic Press, Boston, 2003.
[19] Lee KC and Ou JS, Huang CW, “Angular-Diversity Radar Recognition of Ships by Transformation Based Approaches --- Including Noise Effects,” Progress In Electromagnetics Research-PIER, Vol.72, pp. 145-158, 2007.
[20] Lee KC, Ou JS, and Huang MC, “Underwater Acoustic Localization by Principal Components Analyses Based Probabilistic Approach” , Applied Acoustics, Vol. 70, No. 9, pp. 1168-1174, 2009.
[21] Belhumeour PN, Hespanha JP, and Kriegman DJ, “Eigenfaces vs. Fisherfaces: Recognition Using Class Specific Linear Projection”, IEEE Trans. Pattern Analysis and Machine Intelligence, Vol. 19, No. 7, pp. 711-720, 1997.
[22] Swets DL and Weng JJ, “Using discriminant eigenfeatures for image retrieval”, IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 18, Issue 8, pp. 831 -836, 1996.
[23] Weng JJ, “Crescepton and SHOSLIF: Towards Comprehensive Visual Learning”, Early Visual Learning, S.K. Nayar and T. Poggio, eds., pp. 183-214, Oxford Univ. Press, 1996.
[24] Lee KC and Ou JS, “Radar Target Recognition by Using Linear Discriminant Algorithm on Angular-Diversity RCS,” Journal of Electromagnetic Waves and Applications, Vol. 21, No. 14, pp.2033-2048, 2007.
[25] Lee KC, Wang LT, Ou. JS, and Huang CW, “An Efficient Algorithm for the Radar Recognition of Ships on the Sea Surface”, 2007 International Symposium of MTS/IEEE Oceans, 2007.
[26] Hermus K, Dologlou I, Wambacq P, and Compernolle DV, “Fully adaptive svd-based noise removal for robust speech recognition”, European Conference on Speech Communication and Technology, pp. 1951-1954, 1999.
[27] Jensen SH, Hansen PC, Hansen SD, and Sorensen JA, “Reduction of broad-band noise in speech by truncated qsvd”, IEEE Transactions on Speech and Audio Processing, vol. No. 3, pp. 439-448, 1995.
[28] Konstantinides K, Natarajan B, and Yovanof GS, “Noise estimation and filtering using block-based singular value decomposition”, IEEE Transactions on Image Processing, Vol. 6, Issue 3, pp. 479-483, 1997.
[29] Moon TK and Stirling WC, Mathematical Methods and Algorithms for Signal Processing, Prentice Hall, 2000.
[30] Lee KC, Ou JS, and Fang MC, “Application of SVD Noise-Reduction Technique to PCA based Radar Target Recognition”, Progress In Electromagnetics Research- PIER, Vol.81, pp.447-459, 2008.
[31] Breiman L, Friedman JH, Olshen RA, and Stone CJ, Classification and Regression Trees, Chapman & Hall, Boca Raton, 1984.
[32] Bittencourt HR and Clarke RT, “Use of Classification and Regression Trees (CART) to Classify Remotely-Sensed Digital Images”, 2003 IEEE International Geoscience and Remote Sensing Symposium, 2003.
[33] Deconinck E, Hancock T, Coomans D, Massart DL, and Heyden YV, “Classification of drugs in absorption classes using the classification and regression trees (CART) methodology”, Journal of pharmaceutical and biomedical analysis, Vol. 39, pp. 91-103, 2005.
[34] Yim J, “Introducing a decision tree-based indoor positioning technique”, Expert Systems with Applications, Vol. 34, Issue 2, pp. 1296-1302, 2008.
[35] Gouaillier V and Gagnon L, “Ship silhouette recognition using principal components analysis”, SPIE Proc. Applications of Digital Image Processing XX , Vol. 3164, pp. 59-69, 1997.
[36] Ruck, GT, Barrick DE, Stuart WD, and Krichbaum CK, Radar Cross Section Handbook, Vol. 1, Plenum, New York, 1970.
[37] Crispin JW, Jr. and Maffett AL, “Radar Cross-Section Estimation for Simple Shapes”, Proceeding of the IEEE, Vol. 53, No. 8, pp. 833-848, August 1965.
[38] Paddison FC, Shipley CA, Maffett AL, and Dawson MH, “Radar Cross Section of Ships”, IEEE Transaction on Aerospace and Electronic Systems, Vol. AES-14, pp. 27-34, 1978.
[39] Ewell G and Zehner S, “Bistatic Radar Cross Section of Ship Targets”, IEEE Journal of Oceanic Engineering, Vol. 5, Issue 4, pp. 211-215, 1980.
[40] Tice TE, “An Overview of Radar Cross Section Measurement Techniques”, IEEE Transactions on Instrumentation and Measurement, Vol. 39, Issue 1, pp. 205-207, 1990.
[41] Milne PH, Underwater Acoustic Positioning Systems, Gulf Publishing, Houston, 1983.
[42] Vickery K, “Acoustic positioning systems. A practical overview of current systems”, Autonomous Underwater Vehicles, pp. 5-17, 1998.
[43] Thomas HG, ”GIB buoys: An interface between space and depths of the oceans,” Autonomous Underwater Vehicles, pp. 181-184, 1998.
[44] Raleigh GG and Boros T, “Joint space-time parameter estimation for wireless communication channel”, IEEE Transactions on Signal Processing, Vol. 46, Issue 5, pp. 1333-1343, 1998.
[45] Swindlehurst AL, “Time delay and spatial signature estimation using known asynchronous signals”, IEEE Transactions on Signal Processing, Vol. 46, Issue 2, pp. 449-462, 1998.
[46] van der Veen AJ, Vanderveen MC, and Paulraj A, “Joint angle and delay estimation using shift-invariance techniques”, IEEE Transactions on Signal Processing, Vol. 46, Issue 2, pp. 405-415, 1998.
[47] Wax M and Leshem A, “Joint estimation of time delays and directions of arrival of multiple reflections of a known signal”, IEEE Transactions on Signal Processing, Vol. 45, Issue 10, pp. 2477-2484, 1997.
[48] Zhang Q and Huang J, “Joint estimation of DOA and time-delay in underwater localization”, Proceedings of the IEEE international conference on acoustics, speech, and signal processing, Vol. 5, pp. 2817-2820, 1999.
[49] Fisher RA, “The statistical utilization of multiple measurements”, Annals of Eugenics, Vol. 8, pp. 376–386, 1938.
[50] Wilks SS, Mathematical Statistics, Wiley, New York, 1963.
[51] Rothwell EJ, Chen KM, Nyquist DP, Ross JE, and Bedermeyer R, “A radar target discrimination scheme using the discrete wavelet transform for reduced data storage”, IEEE Transactions on Antennas and Propagation, Vol. 42, Issue 7, pp. 1033-1037, 1994.
[52] Rasmussen JL, Haupt RL, Michael J, and Walker, “RCS feature extraction from simple targets using time-frequency analysis”, SPIE Proc. Radar Processing, Technology, and Applications, Vol. 2845, pp. 66-74, 1996.
[53] Duda RO, Hart PE, and Stork DG., Pattern Classification, 2nd edition, John Wiley & Sons Inc., 2001.
[54] Haykin S, Neural Networks - A Comprehensive Foundation, Second Edition, Prentice Hall, 1999.
[55] Lee KC, Ou JS, Huang MC, and Fang MC, “A Novel Location Estimation Based on Pattern Matching Algorithm in Underwater Environments”, Applied Acoustics, Vol. 70, No. 3, pp. 479-483, 2009.
[56] Youssef M and Agrawala A, “Handling samples correlation in the hours system”, INFOCOM, pp. 1023–1031, 2004.
[57] Lim H, Kung LC, Hou J, and Luo H, “Zero-configuration, robust indoor localization: theory and experimentation”, Proceedings of the IEEE INFOCOM, 2006.
[58] Kim KI, Jung K, and Kim HJ, “Face recognition using kernel principal component analysis”, IEEE Signal Processing Letters, Vol. 9, Issue 2, pp. 40-42, 2002.
[59] Schölkopf B, Smola A, and Müller K, “Kernel principal component analysis”, Advances in Kernel Methods-Support Vector Learning, pp. 327-352, 1999.
[60] X. He and P. Niyogi, “Locality preserving projections”, Proc. 16th Conf. Neural Information Processing Systems, Vancouver, Canada, pp. 585-591, 2003.
[61] X. He, Yan S, Hu Y, Niyogi, P, and Zhang HJ, “Face recognition using Laplacianfaces”, IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 27, Issue 3, pp. 328-340, 2005.
[62] Xu D, Yan S, Tao D, Lin, S and Zhang HJ, “Marginal Fisher Analysis and Its Variants for Human Gait Recognition and Content- Based Image Retrieval”, IEEE Transactions on Image Processing, Vol. 16, Issue 11, pp. 2811-2821, 2007.
[63] Roweis ST and Saul LK, “Nonlinear dimensionality reduction by locally linear embedding”, Science, pp. 2323-2326, 2000.
[64] Tenenbaum J, Silva V, and Langford J, “A Global Geometric Framework for Nonlinear Dimensionality Reduction”, Science, Vol. 290, No. 22, pp. 2319-2323, 2000.