臺灣博碩士論文加值系統

內涵檢索（Content-Based Retrieval）為目前影像檢索技術研究的重心，也已獲得相當多的進展，然而其為低階的特性與人類高階的圖形認知和語意表達存在著先天性的落差，這說明了為什麼現有的圖形檢索系統多數仍以文字檢索（Text-Based）為主，而在可見的未來，語意式（Semantic）文字檢索的技術仍不可或缺。

本研究嘗試在現有的語意式檢索技術中引入所謂AHP（Analytical Hierarchy Process）方法的應用，同時結合源自ICONCLASS的分類技術，使得語意式的檢索能將任一幅圖片視為多項獨立的實體（entities），這實體可能是一個人物、某一形狀、一項活動甚或是一個抽象觀念，這些獨立的實體可作為檢索分類的依據，同時可進一步賦予它加權值（Weighting），檢索出的圖片可依加權值等級順序排列顯示。另外這項應用與內涵檢索，就某一角度而言，存在著某些相似性，未來也許可以扮演縮短高階語意與低階檢索間的差距的橋樑。

為驗證本研究的可行性，研究中設計了一個應用AHP方法的分類檢索測試系統，同時也加入了相似回饋（Relevant Feedback）的功能，使檢索的準確度提高。另外本研究以經濟部智慧財產局現有的商標圖形檢索系統作為研究對照參考系統，最後我們採用準確度（Precision）、召回度（Recall）評估其績效，結果證實是滿意的。

Content-based retrieval is the focus of current image retrieval research and it indeed has obtained a great achievement. The success, nevertheless, is shadowed by the innate discrepancy between machine and human perception to the image. This gives the explanation why the major existed IR system is still text-based. It further supports the possibility that semantic text-based retrieval is dispensable in the future.

This research aims to find a way to shorten the gap by applying the Analytic Hierarchy Process method, which is a well-known decision making process in the field of management, to the text-based classification and retrieval. This approach, integrated with ICONCLASS-like classification system, enables user to extract the features or objects from image semantically. However, it gives weights as well to these subjects analogously to content-based retrieval does. The retrieved images are sorted and displayed according to their similarity weight.

In order to evaluate the feasibility, an experimental system is designed. All the images used come from Taiwan Intellectual Property Office’s Trademark Classification and Retrieval System, which also serves as references to the results from the research. The outcome of relevant feedback and the evaluation of precision and recall are proved with satisfaction.

中文提要----------------------------------------------------i
英文提要----------------------------------------------------iii
誌謝--------------------------------------------------------v
目錄--------------------------------------------------------vi
表目錄------------------------------------------------------viii
圖目錄------------------------------------------------------ix
壹、緒論----------------------------------------------------1
1-1 研究背景與動機--------------------------------------1
1-2 研究方法--------------------------------------------3
1-3 研究貢獻--------------------------------------------4
1-4 論文架構--------------------------------------------5
貳、影像分類與檢索技術現況----------------------------------6
2-1 前言------------------------------------------------6
2-2 文字檢索技術（Text-Based Retrieval）----------------7
2-3 內涵檢索技術（Content-Based Retrieval）-------------8
2-3-1 一般特徵--顏色（Color）-----------------------9
2-3-2 一般特徵─紋理（Texture）---------------------11
2-3-3 一般特徵─形狀（Shape）-----------------------12
2-4 文字檢索與內涵檢索技術優缺點比較--------------------13
2-5 檢索技術發展趨勢--------------------------------------14
參、階層式分析處理程序（AHP）與ICONCLASS簡介---------------16
3-1 前言------------------------------------------------16
3-2 階層式分析處理程序（AHP）---------------------------17
3-2-1 何謂AHP？-------------------------------------17
3-2-2 AHP的使用-------------------------------------17
3-2-3 AHP小結論-------------------------------------20
3-3 ICONCLASS分類系統-----------------------------------29
3-3-1 什麼是ICONCLASS？-----------------------------29
3-3-2 ICONCLASS 特點-------------------------------30
3-3-3 ICONCLASS的分類與編碼------------------------31
肆、研究方法------------------------------------------------35
4-0 前言------------------------------------------------35
4-1 ICONCLASS分類技術的應用----------------------------36
4-1-1 智慧財產局商標圖形檢索系統分類編碼法----------36
4-1-2 智慧財產局商標圖形檢索系統分類編碼法缺點------37
4-2 應用AHP方法於改善語意式路徑編碼分類法的缺點--------38
4-2-1 如何應用AHP方法作圖形路徑分類----------------38
4-2-2 如何應用AHP方法賦予圖形路徑碼加權值----------39
4-2-3 具加權值分類碼的圖形檢索----------------------44
伍、系統實作------------------------------------------------47
5-1 系統架構--------------------------------------------47
5-2 圖形分類模組功能------------------------------------48
5-2-1 抽取大分類特徵項並取得路徑碼------------------48
5-2-2 抽取中分類特徵項並取得路徑碼------------------49
5-2-3 抽取小分類特徵項並取得路徑碼------------------49
5-2-4 賦予路徑碼加權值------------------------------50
5-3 圖形檢索模組功能------------------------------------52
5-3-1 三種不同檢索方式------------------------------52
5-3-2 相似回饋（Relevant Feedback）-----------------53
5-3-2-1 漸進法-----------------------------------53
5-3-2-2 涇渭分明法-------------------------------54
陸、實驗結果討論--------------------------------------------75
6-1速度-------------------------------------------------75
6-2 精準度（Precision）---------------------------------75
6-3 召回度（Recall）------------------------------------78
6-4 相似回饋（Relevant Feedback）-----------------------79
6-5 實驗檢討--------------------------------------------80
柒、結論----------------------------------------------------86
捌、參考文獻------------------------------------------------88
附錄一------------------------------------------------------93

捌、參考文獻1.電信研究雙月刊，第27卷 / 第2期，中華電信研究所， 1997。2.胡曉峰等編著，〝多媒體系統原理與應用〞，儒林圖書公司，1996。3.繆紹綱編著，〝數位影像處理─活用Matalab〞，全華科技，1999。4.Rafael C. Gonzalez & Richard E. Woods, Digital Image Processing, Addison Wesley Publishing Company, 1993.5.Satty & Kearns, Analytical Planning — The Organization of Systems, Pergamon Press Inc., 1985.6.Maggie C.Y.Tam & V.M. Rao Tummala, “An application of the AHP in vendor selection of a telecommunications system,” Elsevier Science Ltd., 2001.7.Yang Jienmei, “Saaty''s AHP with inner dependent hierarchy and the choice of key industries,” Engineering Management Conference, 1994. ''Management in Transition: Engineering a Changing World'', Proceedings of the 1994 IEEE International , Page(s): 274 -279, 1994.8.Abby A. Goodrum, “Image Information Retrieval ：An Overview of Current Research,” Information Science , Volume 3 No2, 2000.9.Yong Rui & Thomas S. Huang & Shih-Fu Chang, “Image Retrieval：Past, Present, and Future, “ International Symposium on Multimedia Information Processing, 1997.10.John M. Zachary, Jr., “An Information Theoretic Approach to Content Based Image Retrieval,” B.S., Louisiana State University, Department of Computer Science Dissertation, September 2000.11.Babu M. Mehtre & Mohan S. Kankanhalli & Wing Foon Lee, “Content-Based Image Retrieval Using a Composite Color-Shape Approch,” Information Processing & Management, Vol. 34, No 1, pp.109-120, 1998.12.Liu Wenyen & Susan Dumais & HongJisang Zhang, “Semi-Automatic Image Annotation, “ Paper from “http://research.microsoft.com/users/marycz/semi-auto-annotation.”13.Sanghoon Sull & Jeongtaek Oh & Sangwook Oh, “Relevance Graph-Based Image Retrieval,” IEEE, pp. 713 -716 vol.2 , 2000.14.Randall M. Rohrer & John L. Sibert & David S. Ebert, “A Shape-based Visual Interface for Text Retrieval,” IEEE pp. 40-46, October, 1999.15.Rohini K. Srihari & Zhongfei Zhang & Aibing Rao, “Intelligent Indexing and Semantic Retrieval of Multimodal Documents,” Kluwer Academic Publisher, Boston, 1999.16.Jae-Chang Shim & Chitra Dorai & Ruud Bolle, “Automatic Text Extraction from Video for Content-Based Annotation and Retrieval,” IBM Technical paper: RC21087 in 1998.17.Virginia E.Ogle & Michael Stonebraker, “Chabot：Retrieval from a Relational Database of Images,” IEEE, pp. 40-48, September, 1995.18.Jianping Fan & Xingquan Zhu & Lide Wu, “Automatic Model-Based Semantic Object Extraction Algorithm,” IEEE Transactions on , pp. 1073-1084, Volume: 11 Issue: 10 , Oct. 2001.19.Burak Ozer & Wayne Wolf & Ali N. Akansu, “A Graph Based Object Description for Information Retrieval in Digital Image and Video Libraries,” paper from “http://www.njcmr.org/~ibo8175/paper1.pdf.”20.Changick Kim and Jenq-Neng Hwang, “Fast and automatic video object segmentation and tracking for content-based applications, “ Circuits and Systems for Video Technology, IEEE Transactions on , Volume: 12 Issue: 2 , Feb. 2002 , Page(s): 122 —129.21.John R. Smith and Shih Fu Chang., “Single Color Extrction and Image Query,” In Proc. IEEE Int. Conf. On Image Proc., 1995.22.John R. Smith and Shih Fu Chang., “Tools and Technique for Color Image Retrieval,” In IS & T/SPIE Proceedings Vol.2670, Storage & Retrieval for Image and Video Databases IV, 1995.23.John R. Smith, “Integrated Spacial and Feture Image Systems：Retrieval, Analysis, and Compression,” Columbia University Department of Computer Science Dissertation, 1997.24.Michael Swan and Dana Ballard, “Color indexing,” International Journal of Computer Vision, 7(1),1991.25.W.Niblack, R. Barber, and et al. “The QBIC project：Querying images by content using color, texture, and shape,” In Proc. SPIE Storage and Retrieval for Image and Video Databases, Feb 1994.26.Markus Striker and Markus Orengo, “Similarity of color images,” In Proc. SPIE Storage and Retrieval for Image and Video Databases, 1995.27.Robert M. Haralick, K. Shanmugam, and Its’hak Dinstein, “Texture features for image classification,” IEEE Trans. On Sys, Man, and Cyb, SMC-3(6),1973.28.Calvin C. Gotlieb and Herbert E. Kreyszing, “Texture descriptors based on co-occurrence matrices,” Computer Vision, Graphics, and Image Processing, 51, 1990.29.Tianhorng Chang and C.-C. Jay Kuo, “Texture analysis and classification with tree-strctured wavelet transform,” IEEE Trans. Image Proc., 2(4):429-441, October 1993.30.K.S. Thyagarajan, Tom Nguyen, and Charles Persons, “A maximum likelihood approach to texture classification using wavelet transform,” In Proc. IEEE Int. Conf. On Image Proc., 1994.31.Anlan Kundu and Jia-Lin Chen, “Texture classification using qmf bank-based subband decomposition,” CVGIP：Graphical Models and Image Processing, 54(5):369-384, September 1992.32.Lin, C.S.,& Hwang, C.L., “New forms of shape invariants from elliptic Fourier descriptors,” Pattern Recognition, 20(5), 535-545, 1987.33.Dubois, S.R., & Glanz, F.H, “An autoregressive model approach to two-deimensional shape classification,” IEEE Trans. On Pattern Analysis and Machine Intelligence, 8, 55-66, 1986.34.Lie, W. N. & Chen, Y. C., “Shape Representation and matching using polar signature,” Paper presented at the International Computer Symposium, Taiwan, 1986.35.Babu M. Mehtre, Mohan S. Kankanhalli and Wing Foon Lee, “Content-Based Image Retrieval Using Composite Color-Shape Approach,” Pergamon, Information Processing & Management, Vol. 34, No. 1, 109-120, 1998.36.ICONCLASS introduction : from “http://www.iconclass.nl/”.37.Heng Tao Shen, Beng Chih Ooi, Kian-Lee Tan, “Finding Semantically Related Images in the WWW,” Paper from Singapore National University Department of Computer Science. 38.James Z. Wang, Yanping Du, “RF*IPF: A Weighting Scheme for Multimedia Information Retrieval,” IEEE 0-7695-1183-X/01, 2001.39.Guojun Lu, Ben Williams, “An Integrated WWW Image Retrieval System, ” Paper from “http://ausweb.scu.edu.au/aw99/papers/lu/paper.html.”40.Lawrence, S. & Giles, L, “Accessibility of information on the web,” Nature, Vol.400,pp.107-109, 1999.41.Jain, R(1993) Workshop report: NSF workshop on visual information management system. in Storage and Retrieval for Image and Video Databases (Niblack, W R and Jain, R C, eds ), Proc SPIE 1908, 198-218.42.Ramesh, N. & Sethi, I.K., “Feature identification as an aid to content based image retrieval,” Paper presented at the SPIE Conference on Storage and Retrieval for Image and Visual Databases III, 1995.43.Ang, Y. H., Li, Z. & Ong, S. H., “Image retrieval based on multidimensional featue properties,” Paper presented at the SPIE Conference on Storage and Retrieval for Image and Visual Databases III, 1995.