近年來面板需求量越來越大，薄膜電晶體液晶顯示器(Thin Film Transistor-Liquid Crystal Display；TFT-LCD)在現代生活中，成為最廣泛的生活必需品，且TFT-LCD 具有體積小、省電、低輻射等優勢；對於TFT-LCD 產品缺陷檢測若依賴人工檢測的方式進行，則會因作業人員的不同及工作狀態而對產品有不同的判斷標準，未來面板尺寸大型化，人工檢測勢必造成極大的挑戰。本研究即在開發一套TFT-LCD 自動光學檢測(Automatic Optical Inspection；AOI)系統，應用影像處理技術判斷面板之端子部與非端子部之瑕疵檢測，檢測項目包含TFT-LCD 邊緣之毛邊(Burr)、破碎(Chipping)、刮痕(Scratches)及汙點(Stains)。首先，將影像中瑕疵目標分割出來，應用改良式對比增長強化法(Enhancement by Modified Contrast Stretching Alorithm)調整對比明暗度，以統計式門檻值(Statistical Threshold)分割前景與背景，接著對獲得的影像進行形態學(Morphology)運算以去除不需要的雜點，並保留完整的瑕疵目標區塊，再對此區塊做數值化處理，獲得其瑕疵的特徵值，由特徵值數據做資料庫，接著使用貝氏分類器(Bayes Classifier)及類神經網路分類器(Neural Network Classifier)分別做瑕疵的分類結果；過程中，利用Sobel運算法及輪廓分布圖來做邊緣的檢測，藉此方法加強瑕疵檢測的準確率。根據分類結果，經過自行設計的檢測系統驗證213 組瑕疵影像，貝氏分類器總辨識率達95.8%，類神經網路分類器總辨識率高達96.2%，驗證所設計的方法適合於TFT-LCD 邊緣瑕疵的檢測系統。

Panel demand is increasing today with thin film transistor-liquid crystal display (TFT-LCD) becoming one of the most extensive and important technological products. TFT-LCD has some advantages such as small size, power saving, and low radiation. Artificial ways to detect TFT-LCD defects result in companies using different standards due to different personnel. The panel size will be increasing in the future, andartificial detection will inevitably cause greater challenges withoutuniform guidelines.
In this study, we developed an automatic optical inspection (AOI) system for TFT-LCD to determine the defect of the panel’s terminal and non-terminal via applying image processing technology. The inspectionitems include burr, chipping, scratches and stains on the TFT-LCD edges.During inspection, we will first separate TFT-LCD defects from the image and use the enhancement by modified contrast stretching algorithm to adjust contrast and shading values. Then, we will perform the statistic threshold method to separate the foreground from the background, and then the system will process with the morphology operator, which can delete unnecessary noise while reserving the complete shape of the defect target. We also processed numerical analysis, where we obtained the characteristics of defects and make a database of the characteristics. Next, we used Bayes classifier and neural network classifier to classify defects.In the process, we used Sobel algorithm and profile distribution to make edge inspection. This method enhanced the accuracy of the rate of defect inspection.
According to our classification results, the automatic optical inspection system investigated 213 groups of the defect image by self-developed inspection system. The Bayes classifier and neuralnetwork classifier total distinguish rates are 95.8% and 96.2%, respectively. Therefore, those rates indicated that this study had successfully developed a set of defect inspection system applicable to the defects on TFT-LCD edges.

1.R. C. Gonzalez and R. E. Woods, “Digital Image Processing,” Addison-Wesley, New York, NY, USA, 1992.
2.J. G. Leu, “Image Contrast Enhancement Based on the Intensities of Edge Pixels,” Graphical Models and Image Processing, Vol. 54, No. 6, pp. 497-506, 1992.
3.C. C. Yang, “Image Enhancement by Modified Contrast-Stretching Manipulation,” Optics & Laser Technology, pp.196-201, 2006.
4.Kundu and J. Zhou, “Combination Median Filter,” IEEE Trans. on Image Processing, Vol. 1, No. 3, pp. 422-429, 1992.
5.P. S. Windyga, “Fast Impulsive Noise Removal,” IEEE Trans. on Image Processing, Vol. 10, No. 1, pp. 173-179, 2001.
6.X. Wang, “Adaptive Multistage Median Filter,” IEEE Transactions on Signal Processing, Vol. 40, No. 4, pp. 1015-1017, 1992.
7.J. Rogowska, K. Preston, and D. Sashin, “Evaluation of Digital Unsharp Masking and Local Contrast Stretching as Applied to Chest Radiographs,” IEEE Transactions on Biomedical Engineering, Vol. 35, No.10, pp.817-827, 1988.
8.M. Sonka, V. Hlavac, and R. Boyle, “Image Processing, Analysis and Machine Vision,” PWS Publishing, Pacific Grove , CA, USA, 1999.
9.S. Kim, H. B. Pyo, S. K. Lee, S. Lee, and S. H. Park, “Digital Image Subtraction of Temporally Sequential Chest Images Byrib Image Elimination,” Proceedings of the 22nd Annual International Conference of the IEEE Engineering in Medicine and Biology Society, Chicago, IL, USA, Vol. 3, pp. 1752-1755, 2000.
10.K. Nakashima, “Hybrid Inspection System for LCD Color Filter Panels,” Instrumentation and Measurement Technology Conference, Hamamatsu, Japan, Vol. 2, pp.689-692, 1994.
11.W. S. Kim, D. M. Kwak, Y. C. Song, D. H. Choi, and K. H. Park, “Detection of Spot-Type Defects on Liquid Crystal Display Modules,” Key Engineering Materials, Vol. 270-273, pp.808-813, 2004.
12.M. Moganti and F. Ercal, “Automatic PCB Inspection Algorithms: a Survey,” Computer Vision and Image Understanding, Vol. 63, No. 2, pp. 287-313, 1996.
13.O. D. Richard, E. H. Peter, and G. S. David, “Pattern Classification,” Second Edition, New York, NY, USA, 2001.
14.T. E. Khwang and D. P. Mital, “An Intelligent Vision System for Inspection of Surface Mounted Devices,” IEEE Region 10 International Conference on TENCON '', Melbourne, Australia, Vol. 2, pp. 804-808, 1992.
15.C. Neubauer, “Intelligent X-Ray Inspection for Quality Control of Solder Joints,” IEEE Transformation on Component, Packing, and Manufacturing Technology – PartC, Vol. 20, No. 2, pp. 111-120, 1997.
16.S. Jagannathon, “Automatic Inspection of Wave Solder Joints Using Neural Network,” Journal of Manufacturing Systems, Vol. 16, No. 6, pp. 389-398, 1997.
17.N. K. Chern, P. A. Neow, H. Marcelo, and J. Ang , “Practical Issues in Pixel-Based Autofocusing for Machine Vision,” Proceedings of the 2001 IEEE International Conference on Robotics & Automation Seoul, Korea, pp. 21-26, 2001.
18.王建智、江行全，「貝氏方法在機器視覺檢測分類中之應用」，第四屆全國品質管理研討會，404-414 頁，1997。
19.鍾瑞賢，應用機器視覺與類神經網路於電機零組件品質之自動檢測-以S鐵心為例，碩士論文，元智大學工業工程研究所，桃園，1999。
20.蔡典霖，應用機械視覺於印刷電路板表面元件之檢測，碩士論文，國立清華大學動力機械工程研究所，新竹，2004。
21.M. A. Telles, “High Performance Borland C++ Builder,” Coriolis Group Books, Albany, N. Y., 1997.
22.余明興等編著，Borland C++ Builder 6程式設計經典，文魁資訊，2002。
23.蔡孟凱等編著，C++ Builder 6 完全攻略，金禾資訊，2005。
24.黃文吉編著，C++ Builder與影像處理，儒林圖書有限公司 ，2002。
25.蕭進松編著，數位影像處理，全華科技圖書股份有限公司，1999。
26.鍾國亮編著，影像處理與電腦視覺第三版，東華書局，台北，2006。
27.I. E. Sobel, “Camera Models and Machine Perception,” Ph.D. thesis, Stanford University, May 1970.
28.D. Mery, “Crossing Line Profile: A New Approach to Detecting Defects in Aluminium Die Casting,” Springer Berlin, 2003.
29.D. Mery and D. Filbert, “Automated Flaw Detection in Aluminum Castings Based on the Tracking of Potential Defects in a Radioscopic Image Sequence,” Robotics and Automation, IEEE Trans., Vol. 18, pp.890-901, 2002.
30.N. Otsu, “A Threshold Selection Method from Gray-Level Histogram,” IEEE Trans. Syst., Man, Cybern., Vol. SMC-8, pp. 62-66, 1978.
31.M. Cherist, J. N. Said, and C. Y. Suen, “A Recursive Thresholding Technique for Image Segmentation,” IEEE Trans. on Image Processing, Vol. 7, No. 6, pp. 918-921, 1998.
32.Y. Du, C. I. Chang, and P. D. Thouin, “Unsupervised Approach to Color Video Thresholding,” Optical Engineering, Vol. 43, No. 2, pp. 282-289, 2004.
33.C. R. Giardina, E. R. Dougherty, “Morphological Methods in Image and Signal Processing,” Prentice-Hall, Upper Saddle River, NJ, USA, 1988.
34.E. R. Dougherty, “An Introduction to Morphological Image Processing,” Bellingham, Wash., USA ,1992.
35.R. M. Haralick, K. Shanmugam, and I. Dinstein, “Textural Features for Image Classification,” IEEE Trans. Systems Man and Cybernetics, Vol. 3, pp. 610-621, 1973.
36.S. T. Bow, “Pattern Recognition and Image Preprocessing,” 2nd ed., New York : Marcel Dekker, 2002.
37.K. Fukunaga, “Introduction to Statistical Pattern Recognition,” 2nd ed., Boston: Academic Press, 1990.
38.Andrew R. Webb, “Statistical Pattern Recognition,” 2nd ed., West Sussex, England; Wiley, 2002.
39.葉怡成，類神經網路模式應用與實作，儒林圖書公司，1999。
40.S. Rajasekaran, “Training-Free Counter Propagation Neural Network for Pattern Recognition of Fabric Defects,” Textile Research Journal,Vol. 67, No. 6, pp. 401-405, 1997.