對大量的藝術與繪畫作品做分類是一件非常繁瑣的工作，而且對繪畫類別的認定也是比較主觀的，本論文提出一個以AdaBoost為訓練方式的繪畫主題搜尋系統，藉由提供一些含有共同繪畫主題的查詢影像，例如肖像畫，並配合回饋的機制，讓使用者可以獲得他想要的繪畫作品。
    自從1990年以來，已經有許多的研究在探討如何創造一個理想的模組來描述一張影像的內容，到目前為止的研究發現，以內容為基礎的影像擷取系統（Contented-Based Image Retrieval）所面臨的主要問題是如何有效的消除存在於低階特徵與高階感知之間的鴻溝。為了找出哪些特徵包含著相似的語意，哪些特徵又可以區分不同的語意，本系統利用了一組不小的特徵集合（4,356個特徵），其中包含了SAD、LEP與OC三種特徵。由於繪畫作品不同於自然影像，同樣主題的畫作可能會呈現出全然不同的色調，所以這三種特徵主要是一些繪畫的紋理與空間上的分布組合關係。在使用者提供了初始查詢後，系統利用AdaBoost Algorithm來從這4,356個特徵中挑選出最關鍵的32個特徵來組合成一個分類器來分類資料庫中的影像。雖然AdaBoost可以很有效率的組合出最後的分類器，但由於一個線上的即時查詢系統並不適合要求使用者提供很多的訓練樣本，所以在此提出一個適合本系統的回饋機制，並將AdaBoost做一些修改以期能更有效的利用使用者的回饋來提升分類的正確率。
    實驗的結果顯示在一個有634張繪畫的資料庫裡查詢「人物畫」時，在經過3次的回饋之後，其效能可以到達Precision rate = 0.71、Recall rate = 0.84、Top 100 precision rate = 0.95。

A content-based painting image retrieval system based on AdaBoost is proposed. By providing query examples which share the same semantic concepts, e.g., portraits, and incorporating with relevance feedback (RF), the user can acquire painting images he desires. 
  Despite the great deal of research work dedicated to the exploration of an ideal descriptor for image content since the early 1990’s, content based image retrieval (CBIR) still is crippled from the well known gap between visual features and semantic concepts. In order to find features shared by similar semantic concepts in painting images and able to distinct dissimilar ones, we propose a large set of 4,356 features (including 3 kinds of features: SAD, LEP, OC) based on texture and spatial arrangement of the painting images. After initial query examples and up to three times of RF, the most critical 32 features out of 4,356 are selected by AdaBoost learning algorithm and form a final classifier. We make use of the characteristic of AdaBoost algorithm that it is very efficient in finding a combination of partial weak classifiers, i.e. features, into a strong one, and thus AdaBoost is very suitable for on-line learning. Experiments show a very satisfying result. In query of “portrait” with three RFs, the system shows an approximate 0.71, 0.84, and 0.95 in Precision, Recall, and Top 100 Precision rates. We will try more combinations of features and apply to a larger data base in the near future.

目錄	i
圖目錄	iii
第1章 緒論	1
1.1 研究動機與目的	1
1.2 繪畫搜尋系統	3
1.3 其他相關研究	6
第2章 系統架構	11
2.1 原始的AdaBoost [17]	12
2.2 修改後的AdaBoost	16
2.3 回饋機制	20
第3章 特徵的選擇	21
3.1 Spatial Angular Distribution (SAD)	22
3.2 Local Edge Pattern (LEP) [20]	25
3.3 Orientational Correlogram (OC)	26
第4章 系統實驗結果與分析	28
4.1 實驗資料庫	28
4.2 系統的執行與操作	29
4.3 效能評估方法	34
4.4 系統的參數	35
4.4.1 T值的測試：	35
4.4.2 ρ值的測試：	37
4.5 特徵的效能	39
4.5.1 影像的前處理	39
4.5.2 SAD的效能	42
4.5.3 LEP的效能	43
4.5.4 OC的效能	43
4.5.5 綜合特徵的效能	43
4.6 關於邊緣影像的門檻值	49
4.7 關於訓練弱分類器	52
4.8 回饋後修改的初始權重	55
4.9 資料庫大小的影響	57
4.10 回饋的影響	59
4.11 其他主題的查詢	60
第5章 結論與未來展望	65
參考文獻	67

圖目錄
圖 2.1  去除極端值（Outliers Removing）	17
圖 3.1 肖像畫	22
圖 3.2  用於SAD的23個區塊	23
圖 3.3 邊緣像素（edge pixels）	24
圖 3.4 用於計算LEP的濾鏡	25
圖 4.1 查詢頁面，讓使用者從其中挑選查詢影像	32
圖 4.2 擷取結果頁面	32
圖 4.3 未被擷取頁面	33
圖 4.4 經過第一次回饋後的擷取結果	33
圖 4.5 T 值的選取	35
圖 4.6 ρ 值的測試	38
圖 4.7 Auto-level	39
圖 4.8 Histogram Equalization	40
圖 4.9 median filter	41
圖 4.10 SAD 前處理	42
圖 4.11 LEP前處理	45
圖 4.12 OC前處理	46
圖 4.13  取3種特徵各自最佳的前處理與合併後的效能	47
圖 4.14  加OC與否的比較	48
圖 4.15 SAD中 取出邊緣像素的門檻值	50
圖 4.16  LEP中 取出邊緣像素的門檻值	51
圖 4.17 微調門檻值	 53
圖 4.18  調整門檻值的效能差異比較	54
圖 4.19 修改初始權重與否的差異	56
圖 4.20 不同的資料庫大小對於效能的影響	57
圖 4.21回饋次數的影響	59
圖 4.22 查詢「靜物畫」	60
圖 4.23 查詢「風景畫」	63

[1]	Hachimura K., “Retrieval of paintings using principal color information,” Pattern Recognition, 1996, Proceedings of the 13th International Conference on Volume 3, 25-29 Aug. 1996 Page(s): 130-134 vol.3.
[2]	J.M. Corridoni, A. Del Bimbo and P. Pala, “Retrieval of paintings using effects induces by color feature,” Content-Based Access of Image and Video Database, 1998. Proceedings, 1998 IEEE International Workshop, 3 Jan. 1998 Page(s): 2–11.
[3]	Y. Isomoto, K. Yoshine, H. Yamasaki and N. Ishii, “Color, shape and impression keyword as attributes of paintings for information retrieval,” Systems, Man, and Cybernetics, 1999. IEEE SMC '99 Conference Proceedings. 1999 IEEE International Conference on Volume 6,  12-15 Oct. 1999 Page(s): 257-262 vol.6.
[4]	P. androutsos, A. Kushki, K.N. Plataniotis and A.N. Venetsanopoulos, “Fuzzy aggregation of palette colors for hybrid querying of fine art image databases,” Digital Signal Processing, 2002. DSP 2002. 2002 14th International Conference on Volume 1, 1-3 July 2002 Page(s): 115-118 vol.1.
[5]	Jing-Nan Liu and Man-Kwan Shan, “A Personalized Fine Arts Painting Style Query System,” 1st Digital Archives Technique Conference, Nangang, Taipei 2002.
[6]	M. Yelizaveta, Chua Tat-Seng, A. Irina and R. Jain, “Representation and Retrieval of Paintings Based on Art History Concepts,” Multimedia and Expo, 2004. ICME '04. 2004 IEEE International Conference on Volume 2, 27-30 June 2004 Page(s):1023 - 1026 Vol.2.
[7]	A. Kushki, P. Androutsos, K. N. Plataniotis and A. N. Venetsanopoulos, “Retrieval of images from artistic repositories using a decision fusion framework,” Image Processing, IEEE Transactions on Volume 13,  Issue 3,  March 2004 Page(s):277 – 292.
[8]	Jeremy S. De Bonet and Paul Viola, “Structure Driven Image Database Retrieval,” Proceedings of the 1997 conference on Advances in neural information processing systems 10.
[9]	Kinh Tieu and Paul Viola, “Boosting Image Retrieval,” Computer Vision and Pattern Recognition. Proceedings of the IEEE Conference on, Volume: 1, 13-15 June 2000.
[10]	L. Brocker, M. Bogen and A. B. Cremers, “Improving the retrieval performance of content-based image retrieval systems: the GIVBAC approach,” Information Visualisation, 2001. Proceedings. Fifth International Conference on 25-27 July 2001 Page(s):659 – 664.
[11]	Guo-Dong Guo, A. K. Jain, Wei-Ying Ma and Hong-Jiang Zhang, “Learning similarity measure for natural image retrieval with relevance feedback,” Neural Networks, IEEE Transactions on Volume 13,  Issue 4,  July 2002 Page(s):811 – 820.
[12]	Ye Lu, Hong-Jiang Zhang, Liu Wenyin and Chunhui Hu, “Joint semantics and feature based image retrieval using relevance feedback,” Multimedia, IEEE Transactions on Volume 5, Issue 3,  Sept. 2003 Page(s):339 – 347.
[13]	Feng-Cheng Chang and Hsueh-Ming Hang, “Content-based image retrieval using both positive and negative feedback,” Multimedia and Expo, 2004. ICME '04. 2004 IEEE International Conference on Volume 3,  27-30 June 2004 Page(s):1887 - 1890 Vol.3.
[14]	A. Kushki, P. Androutsos, K. N. Plataniotis and A. N. Venetsanopoulos, “Query feedback for interactive image retrieval,” Circuits and Systems for Video Technology, IEEE Transactions on Volume 14,  Issue 5,  May 2004 Page(s):644 – 655.
[15]	Feng Jing, Mingjing Li, Hing-Jiang Zhang and Bo Zhang, “Relevance feedback in region-based image retrieval,” Circuits and Systems for Video Technology, IEEE Transactions on Volume 14,  Issue 5,  May 2004 Page(s):672 – 681.
[16]	Xiang Sean Zhou and Thomas S. Huang, “Comparing discriminating transformations and SVM for learning during multimedia retrieval,” International Multimedia Conference; Vol. 9, Proceedings of the ninth ACM international conference on Multimedia, Pages: 137 - 146.
[17]	Y. Freund and R. E. Schapire. “A decision-theoretic generalization of online learning and an application to boosting.” J.Comp. & Sys. Sci., 55(1):119–139, 1997.
[18]	Guo-Dong Guo and Hong-Jiang Zhang, “Boosting for fast face recognition,” Recognition, Analysis, and Tracking of Faces and Gestures in Real-Time Systems, 2001. Proceedings. IEEE ICCV Workshop on 13 July 2001 Page(s):96 – 100.
[19]	Yizheng Cai, “Image Retrieval Using Boosting Algorithm,” http://www.cs.ubc.ca/~yizhengc/ .
[20]	Y.-C. Cheng, S.-Y. Chen, “Image classification using color, texture and regions,” Image and Vision Computing Volume: 21, Issue: 9, September 1, 2003, pp. 759-776.
[21]	T. Ojala, M. Pietikainen, “Unsupervised texture segmentation using features distributions,” Pattern Recognition 32 (1999) 477–486.
[22]	Xipeng Qiu, Zhe Feng, Lide Wu, “Boosting Image Classification Scheme,” IEEE ICME, 2004.
[23]	R.M. Haralick, K. Shanmugan, and I. Dinsten, “Texture Feature For Image Classification”, IEEE Trans. Syst. Man. Cybern., 1973, SMC-3(6): pp. 610 – 621
[24]	J. Huang, S.R. Kumar, M. Mitra, W.J. Zhu, and R. Zabih, “Image Indexing Using Color Correlograms”, International Journal of Computer Vision, 1999, pp. 245-268.
[25]	Auguste Renoir paintings gallery. http://www.renoir.org.yu/default.asp.
[26]	The Vincent van Gogh Gallery. http://www.vangoghgallery.com/index.html.
[27]	Olga's Gallery - Online Art Museum. http://www.abcgallery.com/index.html.
[28]	Taiwan Digital Gallery. http://www.cca.gov.tw/tdg/.
[29]	Academia Sinica. http://www.sinica.edu.tw/main.shtml.
[30]	Academia Sinica twart database. http://ndweb.iis.sinica.edu.tw/twart/System/database_TE/04te_search/index.jsp.