本研究是利用機器視覺的技術自動檢測ITO（Indium Tin Oxide）導電玻璃表面之瑕疵，ITO導電玻璃為觸控面板（Touch panel）及液晶顯示器（LCD）之重要基礎零件，其製程需將可導電的化學物質濺鍍（Sputtering）至玻璃基板上以產生導電之特性，而經濺鍍後之ITO導電玻璃表面不具同質性（Homogeneous）之紋路特徵，由於非同質性（Inhomogeneous）紋路在空間分佈上不具備高度之規律性，因此使用傳統的紋路分析方法並不能夠有效地檢測出瑕疵。
本研究採用非線性擴散（Nonlinear diffusion）的方法對ITO導電玻璃表面進行瑕疵檢測，此方法之特性是利用灰階梯度（Gradient）之資訊對影像中不同梯度的部份進行不同程度之擴散處理，所謂擴散處理是對影像中梯度較低之背景紋路進行平滑，而對梯度較高之瑕疵邊緣抑制平滑，所以此方法運用在ITO導電玻璃表面之檢測上，可以有效地將背景紋路濾除並同時保留住瑕疵原有之輪廓。
非線性擴散在處理的效果上會受到擴散函數及擴散參數的影響，因此本研究主要針對這兩項因素進行探討，並找出適合用於ITO導電玻璃表面檢測之參數，在選定適當的參數下，實驗結果皆能正確地將背景紋路濾除並凸顯瑕疵，驗證了此方法對於具非同質性紋路之ITO導電玻璃表面有良好的檢測效果。

The purpose of this study is to discuss the use of machine vision in automatic defects inspection in inhomogeneous textures. These inhomogeneous textures include materials such as sputtered glass substrates of touch panels. These textures have similarities in some regions while not in the whole image. Therefore, good processing effect cannot be achieved on these textures by using traditional processing methods.
The proposed method is based on the nonlinear diffusion, in which the edge preserving smoothing filter is dependent on image gradient. This method can smooth the edges in textured regions and, in the meantime, preserve the borders between regions. After the nonlinear diffusion process being done, the inhomogeneous region in the original image will have an approximate uniform gray level, and yet the shape of defective region will be distinctly preserved.
The factors that we found affecting diffusion process result include diffusivity function, diffusivity parameter and iteration times. In this paper, we have analyzed the effects of these factors in order to obtain the best effectiveness. Experiments are performed on sputtered glass substrates of touch panels which lead to our conclusion. The experimental results have shown that our proposed method will have promising results for detecting defects in inhomogeneous textures.

中文摘要……………………………………………………………………… I
英文摘要……………………………………………………………………… II
目錄…………………………………………………………………………… III
表目錄………………………………………………………………………… IV
圖目錄………………………………………………………………………… IV
第一章 緒論………………………………………………………………… 1
1.1 研究背景與動機……………………………………………… 1
1.2 研究範圍與目的……………………………………………… 7
1.3 研究貢獻……………………………………………………… 7
1.4 研究方法簡介……………………………………………… 8
1.5 論文架構…………………………………………………… 8
第二章 文獻回顧…………………………………………………………… 10
2.1 表面瑕疵檢測技術之分類…………………………………… 10
2.2 紋路分析……………………………………………………… 11
2.2.1 空間域方法……………………………………………… 12
2.2.2 頻率域方法……………………………………………… 13
2.2.3 傅立葉轉換與影像還原技術於紋路表面之檢測……… 15
2.3 非線性擴散之應用…………………………………………… 16
第三章 研究方法…………………………………………………………… 19
3.1 研究方法流程概述…………………………………………… 19
3.2 非線性擴散簡介…………………………………………… 21
3.3 擴散函數之選擇…………………………………………… 30
3.4 選擇遞減的擴散參數……………………………………… 36
3.5 擴散參數起始值之決定…………………………………… 43
3.6 疊代終止的條件…………………………………………… 46
第四章 實驗結果…………………………………………………………… 52
4.1 系統架構、實驗環境及實驗參數……………………………… 52
4.2 ITO導電玻璃表面之檢測結果………………………………… 54
第五章 結論與建議………………………………………………………… 67
參考文獻…………………………………………………………………… 68
附錄一 應用於結構式紋路與同質性之隨機紋路之檢測結果…………… 72
附錄二 非線性擴散與一般平滑濾波器及傅立葉轉換對非同質性之隨機
紋路處理效果之比較……………………………………… 78
附錄三 改變擴散函數的次方項對檢測結果之影響……………………… 82
附錄四 改變遮罩大小及其係數對檢測結果之影響…………………… 87
附錄五 擴散參數K為定值時，過大或過小的K對檢測效果的影響…… 90
附錄六 光源變動對非線性擴散檢測結果之影響………………………… 93
附錄七 程式說明…………………………………………………………… 96
表目錄
表3-1 本研究採用之擴散函數、擴散參數及其起始值與疊代終止之設定 47
表4-1 非線性擴散疊代次數與時間關係…………………………………… 53
圖目錄
圖1-1 結構式紋路及各種隨機性紋路之影像…………………………… 3
圖1-2 ITO導電玻璃表面之二維傅立葉影像還原處理………………… 5
圖1-3 ITO導電玻璃表面之三維傅立葉影像還原處理………………… 6
圖1-4 ITO導電玻璃表面之非線性擴散處理……………………………… 9
圖3-1 本研究方法之處理流程…………………………………………… 20
圖3-2 線性擴散與非線性擴散濾除影像中雜訊之效果………………… 22
圖3-3 擴散函數與影像灰階梯度之關係圖……………………………… 28
圖3-4 流量函數與影像梯度之關係圖…………………………………… 29
圖3-5 不同擴散函數處理低解析度下之ITO導電玻璃表面之效果I…… 32
圖3-6 不同擴散函數處理低解析度下之ITO導電玻璃表面之效果II…… 33
圖3-7 不同擴散函數處理高解析度下之ITO導電玻璃表面之效果I… 34
圖3-8 不同擴散函數處理高解析度下之ITO導電玻璃表面之效果II…… 35
圖3-9 擴散參數為 之遞減率曲線圖……………………… 39
圖3-10 擴散參數為 之遞減率曲線圖…………………… 40
圖3-11 以不同遞減率的擴散參數進行非線性擴散之結果……………… 41
圖3-12 隨機紋路影像的平均梯度………………………………………… 45
圖3-13 非線性擴散處理紋路影像各次疊代之結果I……………………… 48
圖3-14 非線性擴散處理紋路影像各次疊代之結果II……………………… 49
圖3-15 非線性擴散疊代400次， 之遞減趨勢圖…………… 50
圖3-16 非線性擴散濾波器自動停止疊代之結果………………………… 51
圖4-1 檢測系統架設圖……………………………………………………… 52
圖4-2 無瑕疵之非同質性隨機紋路實驗樣本……………………………… 55
圖4-3 有瑕疵之非同質性隨機紋路實驗樣本……………………………… 57
圖4-4 實驗樣本Normal.01的表面瑕疵檢測結果………………………… 59
圖4-5 實驗樣本Normal.02的表面瑕疵檢測結果………………………… 59
圖4-6 實驗樣本Normal.03的表面瑕疵檢測結果………………………… 59
圖4-7 實驗樣本Normal.04的表面瑕疵檢測結果………………………… 60
圖4-8 實驗樣本Normal.05的表面瑕疵檢測結果………………………… 60
圖4-9 實驗樣本Normal.06的表面瑕疵檢測結果………………………… 60
圖4-10 實驗樣本Normal.07的表面瑕疵檢測結果……………………… 61
圖4-11 實驗樣本Normal.08的表面瑕疵檢測結果……………………… 61
圖4-12 實驗樣本Normal.09的表面瑕疵檢測結果……………………… 61
圖4-13 實驗樣本Normal.10的表面瑕疵檢測結果……………………… 62
圖4-14 實驗樣本Defect.01的表面瑕疵檢測結果……………………… 63
圖4-15 實驗樣本Defect.02的表面瑕疵檢測結果………………………… 63
圖4-16 實驗樣本Defect.03的表面瑕疵檢測結果………………………… 63
圖4-17 實驗樣本Defect.04的表面瑕疵檢測結果………………………… 64
圖4-18 實驗樣本Defect.05的表面瑕疵檢測結果………………………… 64
圖4-19 實驗樣本Defect.06的表面瑕疵檢測結果………………………… 64
圖4-20 實驗樣本Defect.07的表面瑕疵檢測結果………………………… 65
圖4-21 實驗樣本Defect.08的表面瑕疵檢測結果………………………… 65
圖4-22 實驗樣本Defect.09的表面瑕疵檢測結果………………………… 65
圖4-23 實驗樣本Defect.10的表面瑕疵檢測結果………………………… 66
圖4-24 實驗樣本Defect.11的表面瑕疵檢測結果………………………… 66

郭志嘉，2002年，「應用三維傅立葉分析於非同質性隨機紋路表面之瑕疵檢測」，碩士論文，私立元智大學工業工程與管理研究所。
戴祥林，1998年，「以混合式檢測法為基礎作為印刷電路板自動檢測的實現」，碩士論文，國立交通大學電機與控制工程系。
Acton, S. T., J. P. Havlicek and A. C. Bovik, 2001, “Oriented texture completion by AM-FM reaction-diffusion,” IEEE Transactions on Image Processing, Vol.10, pp. 885-896.
Bakalexis, S. A., Y. S. Boutalis and B. G. Mertzios, 2002, “Edge detection and image segmentation based on nonlinear anisotropic diffusion,” IEEE International Conference on Digital Signal Processing, Vol. 2, pp. 1203-1206.
Bakhadyrov, I., M. A. Jafari, T. Fang, A. Safari, S. Danforth and N. Langrana, 1998, “Defect defection in indirect layered manufacturing,” IEEE International Conference on Systems, Man and Cybernetics, Vol. 5, pp. 4251-4256.
Barash, D., 2002, “A fundamental relationship between bilateral filtering, adaptive smoothing, and the nonlinear diffusion equation,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 24, pp. 844-847.
Chan, C. H. and K. H. Pang, 2000, “Fabric defect detection by Fourier analysis,” IEEE Transactions on Industry Application, Vol. 36, pp. 1267-1276.
Chang, Y. J., H. R. Tyan and H. Y. Liao, 1994, “Texture classification using C-matrix and fuzzy Min-Max neural network,” The 2nd National Conference on Fuzzy & Application, pp. 116-121.
Conci, A. and C. B. Proenca, 1998, “A fractal image analysis system for fabric inspection based on a box-counting method,” Computer Networks and ISDN Systems, Vol. 30, pp. 1887-1895.
Conners, R. W., C. W. Mcmillin, K. Lin and R. E. Vasquez-Espinosa, 1983, “Identifying and locating surface defects in wood : Part of an Automated Lumber Processing System,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. PAMI-5, pp. 573-583.
Deng, H. and J. Liu, 2000, “Unsupervised segmentation of textured images using anisotropic diffusion with annealing function,” The First IEEE Pacific-Rim Conference on Multimedia, University of Sydney, Australia, pp. 62-67.
Erturk, S. and T. J. Dennis, 2000, “Image sequence stabilization based on DFT filtering,” IEE Proc.-Vis. Image Signal Process, Vol. 147, pp. 95-102.
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.
Haralick, R. M., K. Shanmugam and I. Dinstein, 1973, “Textural features for image classification,” IEEE Transactions on System, Man, Cybernetics, Vol. SMC-3, pp. 610-621.
Hong, J. J., K. J. Park, and K. G. Kim, 1998, “Parallel processing machine vision system for bare PCB inspection,” Proceeding of SPIE Conference on Automated Inspection and High-Speed Vision Architectures III, pp. 191-201.
Huang, H. and J. H. Wang, 2000, “Anisotropic diffusion for object segmentation,” IEEE Transactions on System, Man, Cybernetics Conference, Vol. 3, pp. 1563-1567.
Kashioka, S. S., K. Kato and M. Ejiri, 1986, “An automatic vision inspection method for a plastic surface based on image partitioning and gray-level histogram,” Systems and Computers in Japan, Vol. 17, pp. 1296-1305.
Kim, S., M. H. Lee and K. B. Woo, 1999, “Wavelet analysis to fabric defects detection in weaving processes,” Proceedings of the IEEE International Symposium on Industrial Electronics, Vol. 3, pp. 1406-1409.
Kumar, A. and G. Pang, 2001, “Identification of surface defects in textured materials using wavelet packets,” IEEE Transactions on Industry Applications Conference, pp. 247-251.
Lambert, G. and F. Bock, 1997, “Wavelet method for texture defect detection,” Proceedings of the IEEE International Conference on Image Processing, Vol. 3, pp. 201-204.
Lee, C. J. and S. D. Wang, 1999, “Fingerprint feature extraction using Gabor filters,” Electronics Letters, Vol. 35, pp. 288 —290.
Moganti, M., F. Ercal, C. H. Dagli and S. Tsunekawa, 1996, “Automatic PCB inspection algorithms: a survey,” Computer Vision and Image Understanding, Vol. 63, No. 2, pp. 287-313.
Nakashima, K., 1994, “Hybrid inspection system for LCD color filter panels,” Instrumentation & Measurement Technology Conference, Hamamatsu, Japan, pp. 689-692.
Perona, P. and J. Malik, 1990, “Scale space and edge detection using anisotropic diffusion,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol.12, pp. 629-639.
Porat, M. and Y. Y. Zeevi, 1988, “Pattern analysis and texture discrimination in the Gabor space,” Proceedings of the 9th International Conference on Pattern Recognition, Rome, Italy, Vol. 2, pp. 700-702.
Ramana, K. V. and B. Ramamoorthy, 1996, “Statistical methods to compare the texture features of machined surfaces,” Pattern Recognition, Vol. 29, pp. 1447-1459.
Sapiro. G. and D. L. Ringach, 1996, “Anisotropic diffusion of multivalued images with applications to color filtering, ” IEEE Transactions on Image Processing, Vol.5, pp. 1582-1586.
Sarraf, H. S. and J. S. Goddard, 1998, “Robust defect segmentation in woven fabrics,” Computer Vision and Pattern Recognition, Vol. 4, pp. 938-944.
Schmitt, L., 1990, “The practical application of grayscale morphology to the inspection of surfaces,” Proceedings of SME Vision ’90, USA, Vol.11, pp. (8-31)-(8-45).
Su, M. S., C. Y. Chen and K. Y. Cheng, 2002, “Smoothing of 3D polygon meshes using non-linear diffusion equations,” Computer Graphics Workshop, Tainan, Taiwan, pp. 1-7.
Torkamani-Azar, F. and K. E. Tait, 1996, “Image recovery using the anisotropic diffusion equation,” IEEE Transactions on Image Processing, Vol.5, pp. 1573-1578.
Tsai, D. M. and C. Y. Hsieh, 1999, “Automated surface inspection for directional textures,” Image and Vision Computing, Vol. 18, pp. 49-62.
Tsai, D. M. and T. Y. Huang, 2003, “Automated surface inspection for statistical textures,” Image and Vision Computing, Vol. 21, pp. 307-323.
Tsai, D. M. and S. K. Wu, 2000, “Automated surface inspection using gabor filters,” International Journal of Advanced Manufacturing Technology , Vol. 16, pp. 474-482.
Tsuji, H., T. Sakatani, Y. Yashima and N. Kobayashi, 2002, “A nonlinear spatio-temporal diffusion and its application to prefiltering in MPEG-4video coding,” Image Processing Conference, New York, NY, Vol. 1, pp. 85-88.
Woods, P. W., C. J. Taylor, D. H. Copper and R. N. Dixon, 1987, “The use of geometric and gray-level models for industrial inspection,” Pattern Recognition Letters, Vol. 5, pp. 11-17.