本論文研究主要是探討液晶面板模組，在製程中所產生之色不均瑕疵 (mura) 的消除；再利用消除色不均瑕疵系統重覆做影像擷取、影像校正、產生燒錄檔、燒錄資料、影像比對的循環校正，並探討液晶面板中的色不均瑕疵圖像資料或製程上是否有相關性，是否可依照相同關係或特性來做批量的處理直接燒錄補償值；例如，影像的亮度比對、生產製程資料比對，來找出共同可接受的燒錄補償值來做色不均瑕疵校正。
液晶面板色不均瑕疵長久以來都是影響畫質均勻性的一項顯著因素，而且現在為了滿足客戶的需求，大尺寸液晶面板製程對整體的畫面均勻性更為顯著，色不均瑕疵的消除也更加重要。本實驗針對液晶面板的特性與製程資料收集，產生一個共用燒錄補償值 (common code)，直接燒錄到液晶面板上的快取記憶體來做色不均瑕疵消除校正，再大量的驗證，確認色不均瑕疵消除的程度，以恰好可辨識強度差 (just noticeable distortion, JND) 數值與水平-垂直資訊互相差異的程度數值 (H V-value)判定，另再確認迦瑪值 (Gamma value) 光學是否有影響。
實驗以取得先前同一批的液晶面板影像約100 ~ 130片資訊，將這些資訊以每的像素灰階做平均值 (mean)、去除離異值 (outlier-mean)、中位數 (median)、眾數 (mode) 換算出共通的特點當作基準共通補償值 (common code)，再以實際給約100片的液晶面板批量驗證處理的情況下，其水準影像水平資訊互相差異的程度數值約為26 ~ 24，垂直影像資訊互相差異的程度數值約為14 ~ 10，為顯著的差異修正。

In Thin-Film Transistor Liquid Crystal Display (TFT-LCD) panel module manufacturing process, how to repair mura defect, make use of the gets rid of the mura defect system (De-Mura) to again do the catch image, Image region of interest (ROI), the creation de-mura repair code and burn to record the de-mura process procedure that the datas, images to compare right of again repair, and inquire into image information or the manufacturing process last relativity of mura defect in TFT-LCD, whether the processing that can do a batch quantity according to same relation or characteristic directly burns to record to compensate the value, for example: ratio to image and production manufacturing process data, calculate the value (common code) to do the de-mura.
Mura defect of TFT-LCD panel for long time for influence the even of quality as an important factor, and now for satisfying the customer's need, big size TFT-LCD panel, the manufacturing process that de-mura is also more important. The experiment characteristic aiming at TFT-LCD panel and manufacturing process data collections produce a common code burn to record in the panel of take a flash IC to do De-Mura, again effect of a great deal of identification confirmation de-mura, the use just noticeable distortion (JND) value and degree number of image information mutual difference of horizontal-vertical direction (H and V values) to judge, another confirm the gamma value optics whether have the influence.
Experiment with the TFT-LCD panel that obtains in times before the same batch image about 100 ~ 130 pcs of information, these information with each of pixel bright degree operation of the use mean, outlier-mean, median and mode, calculate the common code, use 100 pcs of TFT-LCD panel again to in great quantities identify (batch run). The degree number of image information mutual difference of horizontal direction (H value) is about 26 ~ 24, vertical direction (V value) is about 14 ~ 10, this method is obvious improvement.

[1] 游孟潔譯, "何謂液晶顯示&彩色液晶顯示器之構成材料," TFT 彩色液晶顯示器, 全華圖書股份有限公司, 台灣, 2008, Ch.2 & Ch.4.
[2] 田民波著, 林怡欣校訂, "LCD的工作模式及顯示螢幕構成," TFT LCD 面板設計與構裝技術, 五南圖書出版股份有限公司, 台灣, 2010, Ch.6.
[3] 方立德, "TFT液晶顯示面板的Mura瑕疵自動偵測," 國立交通大學電子工程學系電子研究所, 碩士學位論文, 台灣, 2005.
[4] 林曜璋, "彩色濾光片瑕疵特徵選取與分類之研究," 義守大學工業程管理學系, 碩士學位論文, 台灣, 2002.
[5] 鄞毅捷, "以對比為基礎的液晶面板瑕疵偵測系統," 國立交通大學電子工程學系, 碩士學位論文, 台灣, 2009.
[6] L. T. Fang, H. C. Chen, and C. H. Kuo, “Automatic Mura detection system for liquid crystal display panels,” in Proc. of SPIE, San Jose, CA, Jan.15, 2006, vol.6070, pp.143-152.
[7] K. N. Choi, J. Y. Lee, and S. I. Yoo, “Area-mura detection in TFT-LCD panel,” in Proc. of SPIE, Vision Geometry XII, San Jose, CA, Jan.18-19, 2004, vol.5300, pp.151-158.
[8] K. Taniguchi, K. Ueta, and S. Tatsumi, “A detection method for irregular lightness variation of low contrast,” in Proc. IEEE Int. Conf. on Systems, Man and Cybernetics, Hague, Netherlands, Oct.10-13, 2004, pp.6401-6406.
[9] S. I. Baek, W. S. Kim, T. M. Koo, and K. H. Park, “Inspection of defect on LCD panel using polynomial approximation,” in Proc. IEEE Conf. on TENCON, Chiang Mai, Thailand, Nov.21-24, 2004, pp.235-238.
[10] Y. Zhang and J. Zhang, “A fuzzy neural network approach for quantitative evaluation of mura in TFT-LCD,” in Proc. IEEE Int. Conf. on Neural Networks and Brain, Beijing, China, Oct.13-15, 2005, pp.424-427.
[11] C. H. Chou and Y. C. Li, "A perceptually tuned subband image coder based on the meausure of just-noticeable-distortion profile," IEEE. Trans. Circuits and Systems for Video Technology, vol.5, no.6, pp.467-476, 1995.
[12] Y.Mori, R.Yoshitake, and T.Tamura, "Evaluation and discrimination method of mura in liquid crystal displays by just noticeable difference observation," Proceedings of the SPIE, the International Society for Optical Engineering, vol.4902, pp.715-720, 2002.
[13] M. Yumi, T. Kohsei, and T. Satoshi, "Quantitative evaluation of visual performance of liquid crystal displays," Proceedings of the SPIE, Algorithms and Systems for Optical Information Processing IV, vol.4113, pp.242-249, 2000.
[14] 林士凱, 張慈牧, 吳戴仰譯, "AMLCD的製造," 主動式液晶顯示器(AMLCD) 基礎與應用, 全華圖書股份有限公司, 台灣, 2007, Ch.3.
[15] N. Jayant, J. Johnston, and R. Safranek, "Signal compression based on, models of human perception," Proceedings of the IEEE, vol.81, no.10, pp.1385-1442, 1993.
[16] 王珮嘉, "人類視覺檢測在液晶顯示器不均勻現象之探討," 國立清華大學工業工程與工程管理學系, 碩士學位論文, 台灣, 2006.
[17] 陳俊志, "人眼閾值在液晶顯示器不均勻現象之探討," 國立清華大學工業工程與工程管理學系, 碩士學位論文, 台灣, 2005.
[18] K. Taniguchi, K. Ueta, and S. Tatsumi, "A Mura detection based on the least detactable contrast of human vision," Proceedings of the SPIE. Computer Imaging Vision Pattern Recongnition and Graphics, vol.32, pp.1024-1030, 2004.
[19] J. Y. Lee, and S.I. Yoo, "Automatic detection of region-mura defect in TFT-LCD," IEICE Trans. on information and system, vol.E87-D, no.10, pp.2371-2378, Oct. 2004.
[20] 吳富貴, "液晶顯示器影像品質迦瑪曲線自動化校正技術之研究," 國立台灣科技大學自動化及控制研究所, 碩士學位論文, 台灣, 2006.
[21] 蔡明忠, "平面顯示科技實務課程講義," 國立台灣科技大學自動化及控制研究所, 台灣, 2006.
[22] CA-210 色彩分析儀 USER manual 廣柏實業股份有限公司, in http://www.prochem.com.tw/product.asp.
[23] 姜亨潤, "TFT LCD 迦瑪曲線設計與調校之研究," 國立台灣科技大學自動化及控制研究所, 碩士學位論文, 台灣, 2007.
[24] 液晶面板技術研發簡介,友達光電股份有限公司AU Optronics Corp., in http://www.auo.com.tw/技術研發.
[25] Minitab統計分析軟體教學資源, 昊青股份有限公司, in http://www.minitab.com.tw/resource/teaching-resources.php.