本研究主要針對彩色濾光片微觀瑕疵進行辨識與分類，檢測的瑕疵種類分為微粒(Grain)、黑色矩陣破洞(Black Matrix Hole, BMH)、透明導電膜(Indium Tin Oxide, ITO)缺陷、缺邊形(Missing Edge and Shape, MES)、亮點(Highlights)以及粉塵(Particle)等六種。首先，使用正交投影法(Orthogonal Projection)將檢測影像中的每個像素定位，然後利用圖形比對將檢測影像中相似的區塊標記出，再找出所有區塊中與標準模版最相似之區塊，作為新的標準模板，亦稱為適應性模板(Adaptive Template Matching)，再以影像相減法(Image Subtraction)將適應性模板與檢測影像中每個區塊的相同座標之像素相減後所得到的影像，利用管制界限法(Control Limit Law)分割出瑕疵區域，然後利用邏輯運算將此檢測影像周圍未比對到的區塊清除且設定為背景，再以形態學(Morphology)的方法保留瑕疵完整結構，之後再擷取瑕疵灰階值、彩色分量(R、G、B)以及長寬比(Aspect Ratio)此五種特徵值作為分類器之輸入。實務驗證上，本研究所提出之瑕疵檢測方法，透過所開發軟體，最快可於0.154秒完成瑕疵辨識。
在微觀瑕疵的分類上，將擷取之五種特徵值利用倒傳遞類神經(Back-Propagation Neural, BPN)網路以及最小距離分類器(Minimum Distance Classifier, MDC)做為瑕疵之分類決策理論。實務驗證上，本研究利用41個瑕疵做為訓練樣本，再將307個測試樣本之特徵值作為兩者分類器之輸入，以倒傳遞網路做為分類器，其總體辨識率為93.7%；以最小距離分類器，總體辨識率為95.0%，證實本研究所提出的最小距離分類法能夠應用自動化光學檢測技術於彩色濾光片微觀瑕疵分類，成功的提升生產良率以及降低成本。

This study focused on color filter micro defect recognition and classification, and inspected six defects, namely, grain, black matrix hole (BMH), indium tin oxide (ITO) defect, missing edge and shape (MES), and highlights and particle. First, orthogonal projection was applied to locate each pixel in a test image, then an image comparison was conducted to mark similar blocks in the test image. The block that was most similar to the standard template was determined, and was used as the new standard template (or adaptive template matching). Then, image subtraction was applied to subtract the pixels in the same location in each block of the test image from that of the adaptive template matching in order to obtain a new image. The control limit law was employed to divide a defected region, and then, a logic operation was carried out to remove the blocks, which are not compared around the test image, to set as the background. The complete defect structure was retained in a morphology method. Third, five feature values, namely, defect grayscale, R, G, and B color components, and aspect ratio were acquired as classifier input. In practical verification, defect recognition could be completed as fast as 0.154 sec using the software, as developed by this study, on the basis of a defect inspection method.
In micro defect classification, a back-propagation neural (BPN) network and a minimum distance classifier (MDC) served as the defect classification decision theories for the five acquired feature values. In practical verification, this study used 41 defects as training samples, and treated the feature values of 307 test samples as the BPN network classifier inputs. The total recognition rate was 93.7%. When a MDC was used, then the total recognition rate was 95.0%, proving that MDC method, as proposed in this study, is capable of applying automatic optical inspection technology in order to classify color filter micro defects, thus, successfully improving production yield and lowering costs.

1. 王信陽，「TFT-LCD關鍵零組件左右廠商勝負」，光連雙月刊，58期，2005年。
2. 陳鵬帆，「以自適應共振理論網路為基礎建構彩色濾光片微觀瑕疵辨識系統之研究」，國立成功大學，工業與資訊管理學系碩士班，碩士論文，2006年。
3. 蔡培林，「LCD 彩色濾光片的瑕疵擷取與分類」，國立中央大學，資訊工程研究所，碩士論文，2009年。
4. K. Nakashima, “Hybrid Inspection System for LCD Color Filter Panels,” IEEE Instrumentation and Measurement Technology Conference, Vol. 2, pp. 689-692, 1994.
5. C. L. Chang, H. H. Chang, and C. P. Hsu, “An Intelligent Defect Inspection Technique for Color Filter,” IEEE International Conference on Mechatronics, pp. 933-936, 2005.
6. K.B. Lee, M.S. Ko, J.J. Lee, T.M. Koo, and K.H. Park, “Defect detection method for TFT-LCD panel based on saliency map model,” TENCON 2004, IEEE Region 10. Conference, Vol. A, pp. 223-226, 2004.
7. D. M. Tsai, and S. C. Lai, “Defect Detection in Periodically Patterned Surfaces Using Independent Component Analysis,” Pattern Recognition Vol. 41, No. 9, pp. 2812-2832, 2008.

8. 陳威仰，「適應性相減法於週期性紋路之表面瑕疵檢測」，元智大學，工業工程與管理學系，碩士論文，2008年。
9. 鐘宜岑，「應用於動態背景中的移動物體影像之偵測與即時追蹤系統」，國立交通大學，電機學院IC設計產業研發碩士班，碩士論文，2008年。
10. P. Hu, Y. Luo, and C. Li, “Chinese Chess Recognition based on Projection Histogram of Polar Coordinates Image and FFT,” Chinese Conference on Pattern Recognition, IEEE, pp.1-5, 2009.
11. M. Frucci, G. Ramella, and G. S. D. Baja “Using Resolution Pyramids for Watershed Image Segmentation,” Image and Vision Computing, Vol. 25, No. 6, pp. 1021-1031, 2007.
12. D. M. Tsai and C. Y. Hsieh, “Automated Surface Inspection for Directional Textures,” Image and Vision Computing, Vol. 18, No. 1, pp. 49-62, 1999.
13. D. M. Tsai, and C. H. Yang, “A Quantile–Quantile Plot Based Pattern Matching for Defect Detection,” Pattern Recognition Letters, Vol. 26, No. 13, pp. 1986-1962, 2005.
14. S. Kaneko, Y. Satoh, and S. Igarashi, “ Using Selective Correlation Coefficient for Robust Image Registration,” Pattern Recognition, Vol. 36, No. 5, pp. 1165-1173, 2003.
15. D. Lu, P. Mausel, and E. Brondi’zio, and E. Moran, “Change detection techniques,” International Journal of Remote Sensing, Vol. 25, No. 12, pp. 2365-2401, 2004.

16. 林育德，「應用自動化光學檢測技術於彩色濾光片辨識系統之開發與研究」，國立台灣科技大學，自動化及控制研究所，碩士論文，2009年。
17. 陳建堂，「偏光膜外觀瑕疵之影像檢測系統開發」，國立台灣科技大學，高分子系，碩士論文，2008年。
18. C.Y. Chang, C.H. Li, Y.C. Chang, and M.D. Jeng, “Automatic Die Inspection for Post-sawing LED Wafers,” Proceedings of the 2009 IEEE International Conference on Systems, San Antonio, pp. 1615-1620, 2009.
19. G. Acciani, G. Brunetti, and G. Fornarelli, "Application of Neural Networks in Optical Inspection and Classification of Solder Joints in Surface Mount Technology", IEEE Transaction on Industrial Informatics, Vol. 2, 2006.
20. R. A. Zoroofi, H. Taketani, S. Tamura, Y. Sato, and K. Sekiya, “Automated inspection of IC wafer contamination,” Pattern Recognition, Vol. 34, No. 6, pp. 1307-1317, 2001.
21. X. Yang, G. Pang, and N. Yung, “Discriminative training approaches to fabric defect classification based on wavelet transform” Pattern Recognition, vol. 37, no. 5, pp. 889-899, 2004.
22. 顧鴻壽等 編著，「平面面板顯示器基本概論」，高立圖書有限公司，2004年。
23. 莊東漢等 編著，「平面顯示器概論」，高立圖書有限公司，2008年。
24. R. W. Sabnis, “Color Filter Technology for Liquid Crystal Displays,” Vol. 20, No. 3, pp. 118-130, 1999.
25. 鐘國亮 編著，「影像處理與電腦視覺」，東華書局，2006年。
26. Gonzalez Woods, “Digital Image Processing 3/e,” Prentice Hall, New York, 2009.
27. N. Otsu, “A Threshold Selection Method from Gray Level Histogram,” IEEE Transactions on Systems, Man and Cybernetics, SMC-8, pp. 62-66, 1978.
28. 林忠杰 編著，「邏輯設計」，五南圖書出版股份有限公司，2004年。
29. 葉怡成 編著，「類神經網路模式應用與實作」，儒林圖書，2009年。
30. 張斐章等 編著，「類神經網路導論：原理與應用」，滄海書局，2010年。
31. 余明興等 編著，「Borland C++ Builder 6 程式設計經典」，文魁資訊，2002年。
32. 蔡孟凱等 編著，「C++ Builder 6 完全攻略」，金禾資訊，2003年。
33. Euresys s. a., “EasyAccess 6.7.1 eVision User’s Guide,” EureSys Company, 1997-2006.