隨著每年推陳出新高階面板，去年4K2K電視面板全球市場滲透率約33%至35%，而今年可望攀升至45%至50%，即成為主流產品，然而高解析面板中所需的偏光膜材料品質要求也越高，產線檢測能力勢必需要優化，本文檢測以自動光學檢查(Automatic Optical Inspection，簡稱AOI)之”狹縫穿透光學系統”檢測偏光膜壓痕之缺陷，運用田口實驗方法參數及白光干涉儀，探討缺陷變形量與AOI特徵指標相關性，並尋求最佳化檢測之參數。
針對”金字塔”與”壓痕”之白缺陷影響檢出能力最大之因子為狹縫孔徑；縮小狹縫孔徑，能有效檢出白缺陷。而增加光圈與檢查光亮值可有效加強黑缺陷的檢出能力。以最佳化參數所檢驗變形缺陷，白缺陷部分增加310%，而檢測黑缺陷部分增加150%。
由白光干涉儀量測金字塔變形量、尺寸及平均斜率，得知缺陷之變形量與平均斜率均會影響人員與AOI判定，但與尺寸較無相關。實驗數據觀察變形量門檻為3.03µm與平均斜率門檻為0.00236 ，若變形程度超出門檻值則會判定不良品。經由交叉比對發現，變形量與AOI之影像的灰階密度線性相關程度高，R^2最高達0.8567，故可依據AOI灰階密度當作判斷金字塔缺陷變形程度指標。

The high-end display panels have come into the market recently. For example, the global market share of 4k2k TV panels was about 33% to 35% last year, but increases to 45% to 50% this year. That is, they are becoming the mainstream product in the market. However, the high-resolution panels require higher quality of polarizing films which require optimization inspection to provide quality products.
In this thesis, experiments of polarizing film deformation defects with automated Optical Inspection (AOI)-“slit edge light system” were conducted according to the Taguchi’s method for correlating the deformation defects and the AOI characteristic index measured by white light interferometer. Optimized detection parameters were also deduced. The factor that affects the white portion of concave-convex defect is the slit width, reducing the width is more effective; increasing the inspection brightness also effectively enhances the detection ability of black defects. After improvement, the detection of white defects increases by 310%; the detection of black defects increases by 150%. The measurement results of the white light interferometer show that the deformation and average slope of the defects affect personnel and AOI determination, but not the size. The experiment deformation limit (threshold) is 3.03μm and the average slope threshold is 0.00236; both can be used to determine the defective product effectively. Moreover, the amount of deformation is highly correlated with the AOI image with R2 being up to 0.8567.

[1] 林慈婷、關旻宗、王思淋、周文賢，偏光板膜材的回收與應用—PVA去碘再生與應用，工業材料雜誌371期，工研院材化所，2017。
[2] 吳玉祥，TFT-LCD顯示原理製程技術與設備，光機電奈米量測系統科技整合研討會，2006。
[3] 住華科技，光學檢查機教材，2008。
[4] 力特光電產品介紹取自，:http://www.optimax.com.tw/polarizer/chinese/productinf.html
[5] 吳永輝，自動化光學檢測Automated Optical Inspection，台灣電路板協會，桃園市，2013。
[6] 陳亮嘉，光學影像之基礎建立，國立台灣大學自動化光學研究室網站，2015，取自：http://140.112.14.7/~measlab/course/2013。
[7] 林時計工業株式會社.，取自: http://5watts.jp/zdrive/hayashi-light/home/products/cat_use01/
[8] 張樂融、張奕威，白光干涉膜厚測量分析，量測中心，2015。
[9] Cheng, Y., “Mean Shift, Mode Seeking, and Clustering,” IEEE Trans. Pattern Analysis and Machine Intelligence, Vol. 17, No. 8, pp. 790-799, 1995,.
[10] Chen, X., Zhou, Y., Huang, X. and Li, C, “Adaptive bandwidth mean shift object tracking,” IEEE International Conference on Robotics, Automation and Mechatronics”, art. No. 4681484, pp. 1011-1017, 2008.
[11] 李輝煌. "田口方法-品質設計的原理與實務." 高立圖書有限公司, 台北市(2000).
[12] Chen, H. C. and L. T. Fang, “LOG-filter-based inspection of cluster Mura and vertical-band Mura on liquid crystal displays,” Proceedings of SPIE Machine Vision Applications in Industrial Inspection XIII, Vol. 5679, pp. 257-265, 2005.
[13] Comaniciu, D. and Meer, P., “Mean Shift: A Robust Approach Toward Feature Space Analysis,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 24, No. 5, pp. 603-619,2002.
[14]泰洛科技股份有限公司. http://60.250.69.183/newpage/product.cfm?AdminType=ProductClass2&ID=4&ProductClassID=2
[15]Tomoko,Suzuki, Digital Camera Magazine,2017。 取自: https://snapshot.canon-asia.com/taiwan/article/zh/camera-basics-1-aperture
[16] Lee, J. Y. and S. I. Yoo, “Automatic Detection of Region-Mura Defect in TFT-LCD,” IEICE Transactions on Information and Systems, Vol. E87-D, No. 10, pp. 2371-2378, 2004.
[17]陳建文，彩色影像邊緣線萃取之研究，碩士論文，測量工程學系，國立成功大學，2002。
[18] Zhou, H., Schaefer, G. and Shi, C., “A mean shift based fuzzy c-means algorithm for image segmentation,” 30th Annual international IEEE EMBS Conference, art. No. 4649857, pp. 3091-3094,2008.
[19] Pratt, W. K., S. S. Sawkar and K. O’Reilly,Automatic blemish detection in liquid crystal flat panel displays, IS&T/SPIE Symposium on Electronic Imagine, Vol. 3306, pp. 2-13, 1998.
[20] 陳志忠，液晶顯示器的像素點缺陷與亮度均一性之自動化檢測，碩士論文，機械工程研究所，私立中原大學，2001 。
[21] Song, Y. C., D. H. Choi and K. H. Park, “Multiscale Detection of Defect in Thin Film Transistor Liquid Crystal Display Panel,” Japanese Journal of Applied Physics, Vol. 43, No. 8A, pp. 5465-5468,2004.
[22] Jiang, B. C., C. C. Wang, and H. C. Liu, “Liquid crystal display surface uniformity defect inspection using analysis of variable exponentially weighted moving average techniques,” International Journal of Production Research, Vol. 43, pp. 67-80, (2005).
[23] 陳冠位，影像處理及類神經網路於晶圓缺陷分析之應用，碩士論文，資訊工程研究所，國立中央大學，2007。
[24] 郭家成年，「薄膜電晶體液晶顯示器 Mura 瑕疵檢測技術之研發」，碩士論文，自動化科技研究所，國立台北科技大學，2005。
[25] 張瑞顯，「應用線性迴歸診斷法於液晶顯示器 Mura 缺陷自動化檢測之設計與實現」，碩士論文，製造工程研究所，國立成功大學，2004。
[26] 周政衛，「TFT-LCD Mura 缺陷自動化光學檢測」，碩士論文，製造工程研究所，國立成功大學，2005。
[27] 陳思宏年，「彩色濾光片 Mura 瑕疵之自動化檢測」，碩士論文，工業工程與管理研究所，私立義守大學，2005。
[28] 陳建堂，偏光膜外觀瑕疵之影像檢測系統開發，高分子工程系，國立臺灣科技大學，2008。
[29] 黃良華，應用理想解類似度順序偏好法於偏光膜之瑕疵檢測，高分子工程系，國立臺灣科技大學，2004。