透過您的圖書館登入
IP:3.144.250.169
  • 學位論文

應用機器視覺於TFT-LCD陣列電路工程中瑕疵辨識之研究-接觸孔洞光罩瑕疵自動分類

Development of an Automatic Contact Hole Mask Defect Classidication System for the Inline Inspection of TFT-LCD Array Engineering

指導教授 : 劉益宏

摘要


受惠於面板大廠的投資熱潮帶動,市面上的平面顯示器檢測設備近年來快速的崛起,但大多數的檢測設備僅針對已完成之面板進行瑕疵偵測,而最常見的瑕疵為面板顯示不均的情況,可區分為小範圍集中區域的顯示不均(Blob Mura)、大範圍集中區域的顯示不均(Large area Mura),以及具有直線之特徵的顯示不均(Line Mura),輕微可以用雷射進行修補,嚴重則必須報廢,此對面板廠商而言即成本的消耗,而影響面板良率的主因在於面板陣列電路工程的製程方面,因此若能在陣列電路工程中及時發現瑕疵並加以補救,則可提升面板之良率。 而目前LCD的瑕疵辨識仍然仰賴大量人工進行,使用人工的缺點除了檢測效率差之外,其主觀的判斷亦會造成檢測上的問題,故本研究導入機器視覺於TFT-LCD面板之陣列電路工程的瑕疵檢測。與市面上檢查機台僅提供偵測而無分類之功能相作比較,本論文所發展的『接觸孔洞光罩之微影製程瑕疵影像辨識系統』為其貢獻。藉由融合各種數位影像處理、紋理特徵抽取以及類神經網路辨識等方法可針對檢查機台所拍攝的瑕疵影像進行自動即時偵測與分類。此系統主要針對常見的7種瑕疵影像進行分類,以使瑕疵分類達到自動化、即時化,並以此系統來提供於工業界嚴格之需求。 經由實驗測試結果可知,系統輸入的實際照片,『接觸孔洞光罩之微影製程瑕疵影像辨識系統』的辨識率可達96%,顯示系統針對瑕疵影像可精確的分類,並可預防下一批面板經過微影製程時產生同樣的瑕疵,對於面板製程中的檢測及良率提升具有相當程度的幫助。此外,系統辨識瑕疵影像的平均時間約為四秒之內,可快速的分辨瑕疵,以符合工業檢測所需求。

並列摘要


Currently, most equipments used in TFT-LCD manufacturing can only detect defects from panels. For example, some of the defects are blob Mura, large area Mura, and line Mura that are a typical region defect of TFT-LCD. However, Mura inspection is performed in cell or module assembly engineering. If defects can be found and repaired in array engineering in real time, the yield rate of panels would thus be increased, and the cost can be reduced. Most companies set up quality control departments to increase the yield rate of their products. However, in order to reduce the influences of human factors, an automatic inspection system is needed. The purpose of this study is to provide an automatic defect recognition system. In this study, we combine theories of digital image processing techniques, statistic textured feature extraction, and neural network, and propose a “Defect Recognition System for the Lithography Process Inspection in the CH(Contact hole)-Mask”. This system is able to automatically classify seven common defects, providing a real-time automatic defect classification. The experimental results show that, this system achieves a recognition rate up to 96%. This means that this system is able to classify, and prevent the same defects occur again during the lithography process and increase the accuracy of inspection and the yield rate. Moreover the developed system is also able to classify one defect image within 4 second, which means that high-speed defect inspection is achieved.

參考文獻


[22] 林思賢,”應用視覺及資料探勘技術於TFT-LCD陣列電路工程線中瑕疵辨識之研究-源/汲電極光罩瑕疵自動分類系統開發”,私立中原大學機械工程研究所,2006年。
[25] 董志偉、楊燕珠,“利用階層式比對之影像內容檢索,” 大同大學資訊經營研究所,2005年。
[23] 薛伊伶,”應用視覺及資料探勘技術於TFT-LCD 陣列電路工程線中瑕疵辨識之研究-島狀半導體光罩瑕疵自動分類系統開發”,私立中原大學機械工程研究所,2006年。
[11] R. Jain, R. Kasturi, and B. G. Schunck, Machine Vision, McGraw-Hill, New York, 1995.
[1] C. J. Lu and D. M. Tsai, “Automatic defect inspection for LCDs using singular value decomposition,” International Journal of Advanced Manufacturing Technology, vol. 25, no. 1-2, pp. 53-61, 2005.

延伸閱讀