近年來隨著 TFT-LCD 產業應用在大尺寸消費性產品（主要為LCD-TV）成長快速，面板廠商相繼投入更大尺寸的廠房，以提高產能及應用尺寸規格。以面板之瑕疵檢測流程來看，尺寸越大的玻璃基板，檢測所花費時間勢必會更多，而檢測後之瑕疵分類及判斷面板是否需要修補、重製、報廢及直接往下一個製程繼續後製工作，為瑕疵檢測檢測最主要之目的。
本論文主要討論的是在薄膜電晶體(Thin-Film Transistor，TFT)製程的瑕疵檢測，以瑕疵的部分是否落在面板之電路區域上為思考方向，將面板上重複之電路區域，稱之為Cell，劃分為幾個區域，瑕疵所在之區域稱為Cell Zone，因為每個區域內所包含的內容不盡相同，因此在不同區域內若被檢測出有瑕疵存在，其後續處理對策也會隨之不同。本論文以Cell Zone作為瑕疵分類的主要特徵值，加上瑕疵大小的特徵值來作瑕疵分類標準，實作出一套瑕疵分類系統，再以分類結果來判斷面板的等級，提供面板下一步動作是要作修補、重製、報廢或直接往下一個製程的依據。瑕疵自動分類亦可提升以人工方式分類的效率和人為的誤判率。

In recent years, TFT-LCD(Thin-Film Transistor Liquid-Crystal Display) is grown up quickly at the application of consumer product, like LCD-TV. The manufacturer have invested larger plants, to provide more capacity and application sizes. For defect detection process, the greater the size of glass substrate, the detecting time more. After the inspection, the defect classification and panel judgment is decided the next process for repair, rework, scrap or pass to next layer process. This is the main purpose for defect inspection.
The purpose of this research is the defect inspection of Thin-Film Transistor(TFT) process, the defect detection on the regional circuit for thought. Cell is the circuit of panel repeat of the region. It is divided into several regions, the region of the defect known as the Cell Zone. Each region have different contents. In different regions, if detected There are defects, its follow-up measures will also deal with different. This research base on Cell Zone to classify as defects in the main features of value, coupled with the size of defects to be defect classification criteria. And using the results to determine the panel for the next process is to repair, rework, scrap or pass to next layer process. Defects can be automatically classified to artificially enhance the efficiency and classification of human misjudgment rate.

[1]http://linux.tnc.edu.tw/techdoc/perl_intro/
[2]http://tomcat.apache.org/tomcat-5.5-doc/index.html
[3]http://www.postgresql.org/files/documentation/pdf/8.3/postgresql-8.3-A4.pdf
[4]http:// www.netbeans.org/kb/trails/web.html
[5]http://www.w3schools.com/w3c/default.asp
[6]Chern-Sheng Lin, Yo-Chang Liao , Yun-Long Lay, Kun-Chen Lee, Mau-Shiun Yeh, High-speed TFT LCD defect detection system with genetic algorithm, Assembly Automation, vol.28(1), 69-76
[7]Jung-Hun Kim,Suk Ahn,Jae Wook Jeon,Jong-Eun Byun,“A High-Speed High Resolution Vision System for the Inspection of TFT LCD“, Symposium on ISIE 2001 (2001)
[8]Randal L. Schwartz, Tom Phoenix, “Perl 學習手冊”, 二版, 歐萊禮, 2002
[9]洪進財, “SVG網頁程式設計實務”, 初版, 文魁資訊, 2008
[10]鐘國亮, “影像處理與電腦視覺”, 第三版, 東華書局, 2006
[11]蕭進松, “數位影像處理”, 全華科技圖書, 1999
[12]蔡篤銘、林品杰、曾彥馨, “應用獨立成份分析濾波器於背光板與TFT-LCD面板之瑕疵檢測”, 第五屆全國AOI論壇與展覽，交通大學，新竹市, 2005
[13]曾彥馨, “應用機器視覺於TFT 面板之表面瑕疵檢測與分類”, 元智大學工業工程與管理研究所碩士論文, 2004。
[14]廖耀章, “高速TFT LCD瑕疵檢測系統”, 逢甲大學自動控制工程學系碩士論文, 2004, 14-16。
[15]Shewhart, W. A. (1931). Economic Control of Quality of Manufactured Product, New York: D. Van Nostrand
[16]鄭春生，林裕章，蕭華特管制圖輔助法則及警告界限之特性和評估，品質學報，第一卷，第一期，1993。