臺灣博碩士論文加值系統

　　近年來由於高科技產業的製程技術越來越先進，使得半導體製程朝向微小化與精密化發展，在此同時，若是品質管制系統還是仰賴傳統的人工檢測，則不但會增加人工成本，而且檢驗人員容易因長時間工作導致檢測效率降低，進而造成品質檢驗不穩定。面對產品越做越精密的時代來臨，只有自動化檢測才能確保生產線有穩定的品質，因此自動化檢測技術有逐步取代傳統人工檢測的趨勢。本研究的檢測樣本為半導體的晶粒產品，由於晶粒產品為高密度複雜電子組件，因此，必須透過有參考樣板的圖形比對方式才能有效地偵測到瑕疵，而目前使用最普遍的圖形比對方法是傳統的2D相關係數法，但其計算量龐大且穩定度易受環境影響，導致檢測的結果不夠穩定。

　　本研究針對晶粒影像的瑕疵檢測，提出混合式的檢測方法，研究方法分為兩個部分，分別針對晶粒表面的複雜區域及灰階一致性高的區域進行檢測。在第一部分提出一種改良的相關係數法，稱為『以灰階差異為基礎之部分資訊相關係數法』，主要是針對晶粒複雜結構的檢測，利用相關係數法對複雜區域敏銳的瑕疵偵測能力，有效地偵測到瑕疵；而在第二部分，本研究針對晶粒影像中灰階一致性高的區域，單獨進行小波轉換 (Wavelet transform) 檢測。本研究檢測58張待測影像，使用本研究所提出來的混合式檢測方法能夠100%檢測出晶粒影像的瑕疵。此外，本研究亦將以灰階差異為基礎之部分資訊相關係數法應用到不同種類的工業檢測影像，均能有效偵測到瑕疵，因此，透過改良式相關係數的應用，將使相關係數法能有更寬廣的應用。

The template matching has been applied to the machine vision area for defect inspection for many years. Traditional 2D correlation coefficient method is one of the well-known methods of the template matching. It could detect the defects efficiently, especially when the defects are in the complex background; however, it is sensitive to environmental changes such as lighting and shifting, so it often causes false alarms in the uniform areas. The testing samples of this paper are dies that are with complicated components area and uniform area. We proposed the combination methods that are modified correlation coefficient method and wavelet transform method to find the defects out.

In this study, we divided the images of dies into the complicated region and the uniform region. The modified correlation coefficient method is used in the complicated region, and the wavelet transform method is applied to the uniform region. After inspecting all of the samples, this combination method could 100% correctly detect the defects. This research also used the modified correlation coefficient method to inspect different type of the flaws of industrial products, and it also had excellent defect detection ability. Thus this modified correlation coefficient method is not only could be applied to the detection of the dies, but also be used to other industrial products.

中文摘要 I
英文摘要 II
誌謝 III
目錄 IV
表目錄 VI
圖目錄 VII
第一章 緒論 1
1.1 研究背景與動機 1
1.2 研究目的 4
1.3 研究範疇與限制 5
1.4 研究架構 7
第二章　文獻回顧 8
2.1 自動化光學檢測方法 8
2.1.1 參考樣板檢測 9
2.1.2 無參考樣板檢測 11
2.1.3 混合式檢測 12
2.2 相關係數法 13
2.3 離散小波轉換 15
2.4 文獻回顧結論 19
第三章　研究方法 21
3.1 研究方法概述 21
3.2 區分複雜區域與灰階一致性高區域 24
3.3 相關係數法 30
3.4 以灰階差異為基礎之部分資訊相關係數法 34
3.5 相關係數法與以灰階差異為基礎之部分資訊相關係數法的比較 42
3.6 灰階一致性高區域檢測 45
3.7 離散小波轉換法 49
第四章 實驗結果與分析 54
4.1 參數的選擇 54
4.1.1 檢測視窗大小的選取 54
4.1.2 Gary level Threshold的選取 62
4.1.3 個別值管制圖的管制界限 66
4.1.4 Correlation Threshold的選定 69
4.1.5 小波基底函數的選擇 72
4.1.6 Wavelet Threshold的選取 75
4.2以灰階差異為基礎之部分資訊相關係數法的敏感度分析 79
4.2.1 位移的敏感度分析 79
4.2.2 旋轉的敏感度分析 84
4.2.3 光源的敏感度分析 89
4.3 實驗成果分析 94
4.4 以灰階差異為基礎之部分資訊相關係數法的其他應用領域 107
第五章　論文結論與未來研究方向 123
參考文獻 125
附錄 128

彭光裕，1999，「應用電腦視覺技術於表面黏著元件印刷電路板之自動檢測新系統設計及開發」，碩士論文，交通大學工業工程與管理研究所。

戴顯權，2001，「資料壓縮」，紳藍出版社，高雄。

李淑惠，2001，「X-RAY即時影像強化與瑕疵偵測之研究」，碩士論文，私立元智大學工業工程與管理研究所。

林祥璋，2003，「應用共變異矩陣法於IC雷射打印瑕疵檢測之研究」，碩士論文，國立台灣科技大學自動化及控制研究所。

黃泰祥，2004，「具備人臉追蹤跟辨識功能的一個智慧型數位監視系統」，碩士論文，私立中原大學電子工程學系。

謝坤翰、蔡篤銘，2003，「球格陣列(BGA)基板表面瑕疵檢測」，中國工業工程學刊，第二十卷，第2期，第125-138頁。

張真誠、陳同孝、黃國峰，2000，「數位影像處理技術」，松崗電腦圖書資料股份有限公司，台北。

張真誠、陳同孝、黃國峰，2001，「電子影像技術」，松崗電腦圖書資料股份有限公司，台北。

陳璋琪，2003，「應用小波理論於印刷電路板缺點之檢測」，碩士論文，國立成功大學電機工程學系暨研究所。

蔡雅惠，1999，「彩色圖形比對：元件搜尋與瑕疵檢測之應用」，碩士論文，私立元智大學工業工程與管理研究所。

楊榮華，2002，「應用灰階共變異矩陣之多重指標於瑕疵檢測」，碩士論文，私立元智大學工業工程與管理研究所。

楊程翔，2004，「應用分位數散佈圖 (Quantile-Quantile Plat)之相似度指標於瑕疵檢測」，碩士論文，私立元智大學工業工程與管理研究所。

Bently, W. A., 1979, “The inspection: an automatic optical printed circuit board (PCB) inspection,” Proc. Soc. Photo-Optical Instrument. Engrs., Vol. 201, pp. 37-47.

Caprari, R. S., 2000, “Duplicate document detection by template matching,” Image and Vision Computing, Vol. 18, pp. 633-643.

Chaudhuri, B. B. and U. Pal, 1998, “A complete printed Bangla OCR system,” Pattern Recognition, Vol. 31, pp. 531-549.

Chen, H. D., 1994, “Computer vision for industrial inspection through linguistic fuzzy variable inputs,” Control and Instrumentation, Vol. 2, pp. 1349-1353.

Choi, M. S. and W. Y. Kim, 2002, “A novel two stage template matching for rotation and illumination invariance,” Pattern Recognition, Vol. 35, 119-129.

Fernandez, X., 1997, “Template matching of binary targets in grey-scale image: a nonparametric approach,” Pattern Recognition, Vol. 30, pp. 1175-1182.

Gerhardt, L. A., R. P. Kraft, P.D. Hill and S. Neti, 1989, “Automated inspection of sandpaper products and processes using image processing,” Proceedings of SPIE Conference on Automated Inspection and High-speed Vision Architectures Ⅲ, Vol. 1197, pp. 191-201.

Kai, B. and D.H. Uwe, 2001, “Template matching using fast normalized cross correlation,” Proceedings of SPIE-The International Society for Optical Engineering, Vol. 4387, pp. 95-102.

Kaneko, S., Y. Satoh and S. Igarashi, 2003, “Using selective correlation coefficient for robust image registration,”Pattern Recognition, Vol. 36, pp. 1165-173.

Lewis, J. P., 1995, “Fast normalized cross-correlation,” Vision Interface, pp. 120-123.

Lee, D.T., 1978, “A computerized automatic inspection system for complex printed thick film patterns,” SPIE-Applied Electronic Imaging System, Vol. 143, pp. 172-177.

Malamasa, E. N., E. G. Petrakisa, M. Zervakisa, L. Petitb and J. D. Legatb, 2003, “A survey on industrial vision systems, applications and tools,” Image and Vision Computing, Vol. 21, pp. 171-188.

Moganti, M., F. C. Ercal, H. Dagli and S. Tsunekawa, 1996, “Automatic PCB inspection algorithms: a survey,” Computer Vision and Image Understanding, Vol. 63, pp. 287-313.

Parker, J. R., 1997, “Algorithms for Image Processing and Computer Vision,” John Wiley and Sons, Canada.

Petkove, D. and E. B. Hinkle, 1987, “A rule-based system for verifying engineering specifications in industrial visual inspection applications,” IEEE Transactions analysis and Machine Intelligence, Vol. PAMI-9, pp. 306-311.

Pratt, W.K., 1978, “Digital Picture Processing,” Wiley-Interscience, New York.
Sarraf, H. S. and J. S. Goddard, 1998, “Robust defect segmentation in woven fabrics,” IEEE Computer Society Conference on Computer Vision and Pattern Recognition, Santa Barbara, CA, USA, Jun 23-25, pp. 938-944.

Schmitt, L., 1990, “The practical application of grayscale morphology to the inspection of surfaces,” Proceedings of SME Vision ’90, Vol. 11, pp. (8-31)-(8-45).

Shankar, N. G. and Z. W. Zhong, 2005, “Defect detection on semiconductor wafer surfaces,” Microelectronic Engineering, Vol. 77, pp. 337-346.

Shioyama, T. and J. Hamanaka, 1996, “Recognition algorithm for handprinted Chinese characters by 2D-FFT,” International Conference on Pattern Recognition, Kyoto Inst. of Tech., Japan, Aug. 25-29, Vol. 3, pp. 225-229.

Sutton, M. A., M. Cheng, W. H. Peters, Y. J. Chao and S. R. Mchneill, 1986, “Application of an optimized digital correlation method to planar deformation analysis,” Image and Vision Computing, Vol. 4, pp. 143-150.

Tang, Y. Y., B. F. Li., M. Hong, J. Liu, C.H. Leung and C. Y. Suen, 1996, “A novel approach to optical character recognition based on Ring-Project-Wavelet-Fractal signature,” Proceedings of ICPR, pp. 325-329.

Tang, Y. Y., H. D. Cheng and C. Y. Suen, 1991, “Transformation-Ring-Projection (TRP) algorithm and its VLSI implementation,” International Journal of Pattern Recognition and Artificial Intelligence, Vol. 5, pp. 25-56.

Tsai, D. M., C. T. Lin and J. F. Chen, 2003, “The evaluation of normalized cross correlations for defect detection,” Pattern Recognition Letters, Vol. 24, pp. 2525-2535.

Tsai, D. M. and C.T. Lin, 2003, “Fast normalized cross correlation for defect detection,” Pattern Recognition Letters, Vol. 24, pp. 2625-2631.

Uenohara, M. and T. Kanade, 1997, “Use of fourier and Karhunen-Loeve decomposition for fast pattern matching with a large set of templates,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 19, pp. 891-898.