本論文中，我們設計了一個圖像自動化加註文字系統。此系統的設計目的，在於能自動的幫助使用者給予未知圖像符合語意的加註文字，藉此以提高圖像檢索及圖像分類系統的準確性。

    本系統是採用圖像低階特徵與語意特徵的結合來給予圖像適當的加註文字。一開始先將所有從網際網路取得的圖像資料，經由過濾得到關鍵字集合作為語意特徵。接下來將所有圖像切割成相同大小的子圖像集合，並利用離散小波轉換及色彩直方圖的方式分別取得材質(Texture)及顏色(Color)上的特徵然後合併成單一特徵以作為圖像的低階特徵。完成此一步驟後，我們利用低階特徵將所有子圖像及其關鍵字分群，這樣可以得到以低階特徵為中心概念的關鍵字群組，並統計群組中每一關鍵字出現的機率。完成上述訓練過程之後，將欲加註文字的圖像以相同方式取得子圖像低階特徵集合，與群組中心點作相似度計算後取得該群組的關鍵字資訊，最後藉由統計其關鍵字的機率來取得具代表性的關鍵字作為其加註文字。

    在本論文中所提出的加註文字方法，在取得語意特徵時考慮了低階特徵的特性，所以最後所得到的加註文字將會與圖像內容有一定的相關程度，意即將會更貼近使用者認知的意涵。所以利用此系統所得到的加註文字用於圖像檢索系統及分類上將會有助於準確性的提升。

In this following proposal, we have designed an automatic image annotation system, using primarily the low level features and other semantic characteristics of the image’s color and texture to create annotations for the images. In this system, we use the Internet to obtain image information as a reference for the process. 
  
Throughout the process, the keywords obtained from the image’s original webpage content are collected as defining features, then all of the images are cropped into the same size and collected. Using wavelet transformation and color histogram, the texture and color features are collected and combined into one single feature, which is the image’s low level feature. Lastly, using the low level feature to allocate the images and keywords, a keyword group which has low-level features as its central concept is created and the images will have appropriate annotations. The purpose of such a design for the system is to automatically assist the user in annotating unknown images, and to increase the accuracy of image searching and image grouping systems.

目錄
第一章 緒論	1
1.1 研究動機與目的	1
1.2 研究內容	4
1.3 研究內容大綱	6
第二章 背景知識與相關研究	7
2.1 資訊檢索模式	7
2.1.1 二元模式(Boolean Model)	7
2.1.2 向量空間模式(Vector Space Model)	8
2.1.3 機率模式(Probabilistic Model)	11
2.2 色彩空間模式	12
RGB色彩空間	13
YCbCr色彩空間	14
2.3 離散小波轉換	14
2.4 色彩直方圖	18
第三章 系統架構	21
3.1 系統架構	21
3.2圖像訓練系統(TRAINING SYSTEM)	23
3.2.1 資料蒐集	23
3.2.2 關鍵字擷取	23
3.2.3 低階特徵擷取	25
3.2.4 分群	30
3.2.5 機率統計	31
3.3加註文字系統(ANNOTATION SYSTEM)	33
第四章 實作與討論	35
4.1實作系統環境:	35
4.2 實作介面介紹	36
4.3 實驗結果	37
4.4實驗結果討論	38
第五章 結論與未來研究方向	40
5.1結論	40
5.2未來研究方向	41
參考文獻	42
附錄	44


圖目錄
圖2.1	THE COSINE OF Ө IS ADOPTED AS  	10
圖2.2 第一次水平切割示意圖	16
圖2.3 第一次垂直切割示意圖	17
圖2.4 三階離散小波轉換示意圖	18
圖2.5色彩直方圖說明實例	20
圖3.1系統架構圖	21
圖3.2圖像前處理流程	26
圖3.3 一階小波轉換示意圖	27
圖3.4 材質特徵擷取流程示意圖	28
圖3.5 分群示意圖	31
圖3.6 機率統計流程示意圖	32
圖3.7 加註文字系統流程示意圖	34
圖4.1 加註文字系統的實作介面圖	36
圖4.2 實驗結果(1)	37
圖4.3實驗結果(2)	38
圖4.4 訓練資料中屬於SHARK類別的圖像	39

[1] Yu-Tao Huang, “Integrating Semantics and Low Level Feathers 
 for the Design of Image Retrieval Systems”, Department of Computer Science and Information Engineering of Tamkang University, 2005.
[2] R. Baeza-Yates and B. Ribeiro-Neto,“Moderm Information Retrieval”, Addison Wesley Longman, Inc, May 1999.
[3] Kwiatkowska, M., Norman, G., and Parker, D. PRISM 2.0: A tool for probabilistic model checking. In Proc. 1st International Conference on Quantitative Evaluation of Systems (QEST’04), pages 322-323, IEEE Computer Society Press. September 2004.
[4] Q. Iqbal, J. Aggarwal, “CIRES: A System for Content-Based Retrieval in Digital Image Libraries”, Proc. International Conference on Control, Automation, Robotics and Vision, pp. 205–210, Singapore, Dec. 2002.
[5] Y. Mori, H. Takahashi & R. Oka. Image-to-word transformation based on dividing and vector quantizing images with words. First International Workshop on multimedia Intelligent Storage and Retrieval Management ,1999 
[6] James Z.Wang, “Integrated Region-Based Image Retrieval", Boston, Kluwer Academic Publishers,2001
[7] Bimbo,A.D.,Visual Information Retriveal,Morgan Kaufmann,Inc.,San Francisco,California,1999
[8] http://en.wikipedia.org/wiki/YCbCr
[9] 陳同孝，張真誠，黃國峰，數位影像處理技術，初版，松崗電腦圖書資料股份有限公司，臺北市，2001 
[10] Craizer, M., Silva, E. A. B. D. and Ramos, E. G., “Convergent algorithms fpr successive approximation vector quantization with applications to wavelet image compression,” IEE Proceedings-Vision Image and Signal Processing, Vol. 146, No.3, pp. 159-164, Jun. 1999.
[11] Dumais, S., J. Platt, D. Heckerman and M.Sahami, Inductive  learning algorithms and representations for text categorization, Proceedings of the 7th International Conference on Information and Knowledge Management, Bethesda, MD, pp. 148-155, 1998.
[12] J. Yan, L. Wenyin,H. Zhang, Y. Zhang, “Thesaurus-aided Approach for Image Browsing and Retrieval”, International Conference on Multimedia and Expo(ICME2001), Japan, 2001.
[13] Wang, H. J. and Huo, C. C. J., “A multi-threshold wavelet coder(MTEC) for high fidelity image compression,” IEEE proceedings International Conference on Image Processing, Vol.1 , pp.652-655, 1997.
[14] Zhong, J. M., Leung, C. H. and Tang, Y. Y., “Wavelet image coding based on significance extraction using morphological operation,” IEE Proceedings Image and Signal Processing, Vol. 146, No. 4, pp.206-210, Aug. 1999.
[15] Sitaram Bhagavathy, Kapil Chhabra, “A Wavelet-based Image Retrieval System,” An ECE 278A Project Report.
[16] Christopher D. Manming, Hinrich Schutze(2000), “Foundations of Statistical Natural Language Processing”, 3rd ed, pp539-543
[17] W. Niblack, R. Barber, and et al. “The QBIC project: Query images by content using color, texture and shape”, Proc. SPIE Storage and Retrieval for Image and Video Databases, Feb, 1994.