由於印刷電路板可能會由下述的瑕疵如髮線(hairline)、針孔(pin-hole)、錯的孔徑、斷線、開路、誤接點、凸刺(spur)等等。因此，如何檢測出印刷電路板瑕疵，顯然是非常重要的技術。

此外，利用傳統自動檢測裝置(automatic test equipment)在檢測已銲好電子元件的印刷電路板時仍有許多限制。而紅外線檢測系統具有影像擷取快速、安裝容易及無接觸問題等優點，最近十幾年以來已廣泛地應用在印刷電路板檢視。

檢視印刷電路板時影像分類(image segmentation)是重要的影像前處理程序。在本論文中，我們提出線性量化灰階值的非監督式神經網路(簡稱GLVQ)，GLVQ以影像的灰階值之像素統計資料(histogram)為輸入，所以輸入資料量僅與影像的灰階值範圍有關而與影像大小無關。由實驗顯示，利用GLVQ將銅箔從電路板影像分離出來的速度高於傳統的線性量化神經網路(LVQ)。

在應用次模型層電路板檢視技術(subpattern level inspection method)時，萃取電路板的銅箔模型是不可或缺的程序。傳統上，以視窗為基礎的電路板銅箔模型萃取技術，可簡化後續偵測瑕疵之處理程序。此模型萃取技術以最大種子視窗萃取技術(簡稱LSWE)最為典型。然而，LSWE仍會遺漏部分銅箔(即無法萃取出所有銅箔模型，在本論文中稱此為不完整覆蓋問題)且對整張影像中各個銅箔區塊面積間的差異值有所限制。在本論文的第三章，我們提出最小種子視窗萃取技術(簡稱SSWE)，在萃取銅箔模型時不會限制各個銅箔面積間的差異值。

更進一步地，我們在本論文的第四章提出植基於輪廓的視窗萃取技術(簡稱CBWE)，既可以克服前述不完整覆蓋問題，也不會限制各個銅箔面積間的差異值。

在第五章我們提出植基於向量量化的電路板熱影像分析技術(簡稱VBCT)。應用VBCT時會先將功能正常的電路板熱影像(在論文中稱此為黃金熱影像)編碼並產生電碼書(code book)。由於VBCT以電碼書取代黃金熱影像，所以可以減少儲存黃金熱影像所需要的記憶空間。在與待測電路板的熱影像比對時，是依我們在第五章所提出的適應性閥值準則(adaptive threshold criterion)與前述電碼書比對。由實驗顯示，VBCT可以標示出工作不正常的電路元件。

Bare PCB (Printed Circuit Board) may have defects such as hairline, pin-hole, wrong size hole, breakout, open circuit, mis-contact, spur, mission feature, etc, and it is essential to detect these defeats effectively.

Conventional automatic test equipment (ATE) has some limitations and drawbacks in testing the produced PCB which has been attached, inserted, and soldered the electronic components. The thermal imaging diagnostic system has the advantages of no contact problem, rapid image acquisition, easy operation, and simple testing reconfiguration. Therefore, thermal imaging approach has been widely applied in the past decade to diagnose faults on PCB.

Image segmentation plays an important role in the inspection of bare PCB. In this thesis, a novel Gray-level-Based LVQ Neural Network (GLVQ) with an unsupervised learning scheme is proposed for PCB inspection. The GLVQ architecture is as simple as that for the conventional LVQ. The input data for the GLVQ network are obtained from the occurrences of gray levels in a PCB image. From the experimental results, it is shown that the proposed GLVQ is very effective in PCB inspection.

Pattern extraction is an indispensable step in bare PCB inspection and plays an important role in automatic inspection system design. A good approach for pattern definition and extraction will make the following PCB diagnosis easy and efficient. The window-based technique has great potential in PCB patterns extraction due to its simplicity. The conventional window-based pattern extraction methods, such as Large Seeds Window Extraction method (LSWE), have the problems of losing some useful copper traces (called as non-cover problem in this thesis) and limiting the variation of the trace’s size. In this thesis in chapter 3, we proposed a new method to extract the basic geometric pattern based on small seed window (SSWE for short). SSWE can extract the traces’ pattern without limiting their size. The experimental results show that the feasibility and effectiveness of the proposed method SSWE can extract the traces’ pattern without limiting their size.

Furthermore, in order to solve the non-cover problem, in chapter 4, we proposed Contour Based Window Extraction (CBWE) algorithm for improvement. From the experimental results, the proposed CBWE algorithm is demonstrated to be very effective in basic pattern extraction from bare PCB image analysis.

To analyze the thermal image of PCB for faults detection, in chapter 5, we have proposed a novel vector quantization (VQ) based approach, which can reduce the memory size, for thermal image analysis of PCB diagnosis. Appling the proposed VQ-based approach, the gold thermal image is coded into a codebook and compared with the board under test (BUT) to identify the image blocks with faults, instead of the whole thermal image. In addition, an adaptive threshold criterion is proposed to improve the detection sensitivity. From the experimental results, this proposed method is demonstrated to be very effective in abnormal functional block identification for PCB based on thermal image.

REFERENCES
[1] A. R. Rao, “Future directions in industrial machine vision: A case study of semiconductor manufacturing applications,” Image and Vision Computing, vol. 14, pp. 3-19, 1996.
[2] H. H. Loh and M. S. Lu, “Printed circuit board inspection using image analysis,” IEEE Trans. Industry Applications, vol. 35, no. 2, 1999.
[3] B. Raghunathan and S.T. Acton, “A content based retrieval engine for circuit board inspection,” Proc. IEEE/ICIP Int. Conf. Image Processing, vol. 1, pp. 104-108, 1999.
[4] T.-H. Cho, R. W. Conners, and P. A. Araman, “A comparison of rule-based, k-nearest neighbor, and neural net classifiers for automated industrial inspection,” Proc. IEEE/ACM Int. Conf. Developing and Managing Expert System Programs, pp. 202-209, 1991.
[5] M. Moganti and F. Ercal, “Automatic PCB inspection algorithms: A survey,” Computer Vision and Image Understanding, vol. 63, no. 2, pp. 287-313, March 1996.
[6] R. T. Chin and C. A. Harlow, “Automated visual inspection: a survey,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 4, no. 6, pp. 557-573, 1982.
[7] R. T. Chin, “Automated visual inspection: 1981 to 1987,” Comput. Vision Graph. Image Process., vol. 41, pp. 346-381, 1988.
[8] W. Y. Wu, M. J. J. Wang, and C. M. Liu, “Automated inspection of printed circuit boards through machine vision,” Computer in Industry, vol. 28, pp. 103-111, 1996.
[9] A. M. Darwish and A. K. Jain, “A rule based approach for visual pattern inspection,” IEEE Trans. Patt. Anal. Mach. Intell., vol. 10, no. 1, pp. 56-68, Jan. 1988.
[10] S. H. OĞUZ and L. Onural, “An automated system for design-rule-based visual inspection of printed circuit boards,” Proc. IEEE Int. Conf. Robotics and Automation, vol. 3, pp. 2696-2701, 1991.
[11] P. Szolgay and K. Tomordi, “Optical detection of breaks and short circuits on the layouts of printed circuit boards using CNN,” Proc. IEEE/CNNA, International Workshop on Cellular Neural Networks and their applications, pp. 87-92, 1996.
[12] H. Tsunekawa, “Latest image evaluation systems aid efforts for product quality,” Journal of Electronic Engineering, vol. 29, no. 306, pp. 72-77, 1992.
[13] M. Moganti and F. Ercal, “A subpattern level inspection system for printed circuit boards,” Computer Vision and Image Understanding, vol. 70, no. 1, April, pp. 51-62, 1998.
[14] M. H. Tatibana and R. de A. Lotufo, “Novel automatic PCB inspection technique based on connectivity,” Proc. X Brazilian Symposium on Computer Graphics and Image Processing, pp. 187-194, 1997.
[15] J. J. Hong, K. J. Park, and K. G. Kim, “Parallel processing machine vision system for bare PCB inspection,” Proc. 24th Annu. Conf. IEEE Industrial Electronics Society, vol. 3, pp. 1346-1350, 1998.
[16] J. H. Koo and S. I. Yoo, “A structural matching for two-dimensional visual pattern inspection,” Proc. IEEE Int. Conf. Systems, Man, & Cybernetics, vol. 5, pp. 4429-4434, 1998.
[17] M. Moganti and F. Ercal, “Segmentation of printed circuit board images into basic patterns,” Computer Vision and Image Understanding, vol. 70, no. 1, April, pp. 74-86, 1998.
[18] M. T. Hagan, H. B. Demuth, and M. Beale, Neural Network Design, PWS Publishing Company, 1996.
[19] H. F. Spence, D. P. Burris, “An artificial neural network printed circuit board diagnostic system based on infrared energy emissions,” Proc. IEEE Systems Readiness Technology Conference, Anaheim, CA USA, 24-26 Sept. 1991, pp. 41-45, 1991.
[20] P. Amy and J. M. J. Walter, “Image processing, a new diagnostic tool,” IEEE Aerospace and Electronics Systems Magazine, vol. 9, Issue: 3 , pp. 34-37, March 1994.
[21] S. C. Yao and K. E. Su, “Thermal analysis of a two-dimensional electronic board,” Proc. I-THERM, Los Angeles, CA USA, 11-13 May 1988, pp. 79-84, 1988.
[22] S. Witzman, D. Newport, and T. Nicoletta, “Free convection air cooling of the electronic equipment: still a miracle waiting to be explored and explained,” Proc. 5th Annual IEEE SEMI-THERM V., San Diego, CA USA, 7-9 Feb. 1989, pp. 98-103, 1989.
[23] W. M. Foster, “Thermal verification testing of commercial printed-circuit boards for spaceflight,” Proc. Annual Reliability and Maintainability Symposium, Las Vegas, USA, pp. 189-195, Jan. 1992.
[24] L. G. Allred and G. E. Kelly, “A system for fault diagnosis in electronic circuits using thermal imaging,” Proc. IEEE Systems Readiness Technology Conference, Dayton, USA, pp. 455-458, Sept. 1992.
[25] B. Linnander, “When it’s too hot to touch use infrared thermography,” IEEE Circuits and Devices Magazine, Vol. 9, no. 4, pp. 35-37, July 1993.
[26] D. H. Lee, “Thermal analysis of integrated-circuit chips using thermographic imaging techniques,” IEEE Trans. Instrum. & Meas., vol. 43, no. 6, pp. 824-829, Dec. 1994.
[27] S. A. Merryman and R. M. Nelms, “Diagnostic technique for power systems utilizing infrared thermal imaging,” IEEE Trans. Indust. Electron., vol. 42, no. 6, pp.615-628, Dec. 1995.
[28] D. Washburn and M. Nguyen, “Commercial off the shelf (COTS)/ non developmental item (NDI) testing for department of defense using infra-red technology,” Proc, Autotestcon Systems Readiness: Test Technology for the 21st Century, Atlanta, USA, pp. 584-587, Aug. 1995.
[29] K. Azar and J. E. Graebner, “Experimental determination of thermal conductivity of printed wiring boards,” Proc. 12th Annu. IEEE SEMI-THERM XII, Austin, USA, pp. 169-182, March 1996.
[30] L. G. Allred and T. R. Howard, “Thermal imaging is the sole basis for repairing circuit cards in the F-16 flight control panel,” Proc. Autotestcon Test Technology and Commercialization, Dayton, USA, pp. 418-424, Sept. 1996.
[31] L. G. Allred and J. P. Harames, “Guidelines for successful implementation of infrared thermography for repairing electronic circuit cards,” Proc. Autotestcon, Test Technology and Commercialization, Dayton, USA, pp. 410-417, Sept. 1996.
[32] P. J. Moore and F. Harscoet, “Low cost thermal imaging for power systems applications using a conventional CCD camera,” Proc. EMPD, vol. 2, Singapore, pp. 589-594, March 1998.
[33] L. G. Allred, “Using infrared thermography to detect age degradation in EPROM chips,” Proc. Autotestcon Systems Readiness Technology Conference, Salt Lake City, USA, pp. 546-551, Aug. 1998.
[34] J. S. Lin, K. S. Cheng, and C. W. Mao, “ A fuzzy Hopfield neural network for medical image segmentation,” IEEE Trans. Nuclear Science, vol. 43, no. 4, pp. 2389-2398, August, 1996.
[35] K. Tomordi, P. Foldesy, and P. Szolgay, “ On the chip implementation of analogic algorithms for optical detection of some layout errors of printed circuit boards,” Proc. 5th Int. Conf. Cellular Neural Networks and Their Applications Proceedings, pp.162-168, 1998.
[36] S. Y. Huang, K. S. Cheng, and C. W. Mao, “Basic patterns extraction for printed circuit board images analysis,” Int. Conf. Computer Vision, Pattern Recognition & Image Processing, pp. 769-772, March 2002.
[37] A. R. Weeks, Jr., Fundamentals of Electronic Image Processing, 1st ed., IEEE Press, 1996, pp. 34-37.
[38] M. P. Dubuisson and A. K. Jain, “ A modified Hausdorff distance for object matching,” Proc. of the 12th IAPR Int. Conf. on Computer Vision & Image Processing, vol. 1, pp. 566 –568, Oct. 1994.
[39] F. P. Incropera and D. P. DeWitt, Fundamentals of Heat Transfer, New York: Wiley, 1981.
[40] X. P. V. Maldague, Infrared Methodology and Technology, Gordon and Breach Science Publishers, 1994.
[41] R. M. Gray, “Vector quantization,” IEEE ASSP Magazine, vol. 1, pp. 4-29, 1984.
[42] Y. Linde, A. Buzo, and R. M. Gray, “An algorithm for vector quantizer design,” IEEE Trans. Commun., vol. COM-28, pp. 85-94, Jan. 1980.
[43] A. Gersho, and R. M. Gray, Vector Quantization and Signal Compression, Kluwer Academic Publishers, Norwell, Ma, 1992.
[44] J. S. Lin, S. H. Liu, and C. Y. Lin, “The application of fuzzy Hopfield neural network to design better codebook for image vector quantization,” IEICE Trans. Fundamentals, vol. E81-A, no. 8, pp. 1645-1651, 1998.
[45] S. H. Liu and J. S. Lin, “A compensated fuzzy Hopfield neural network for codebook design in vector quantization,” Int. J. Patt. Recog. & Artif. Intell., vol. 14, no. 8, pp. 1067-1079, Dec. 2000.
[46] S. W. Ra and J. K. Kim, “A fast mean-distance-ordered partial codebook search algorithm for image vector quantization,” IEEE Trans. Circuits and Systems - Ⅱ: Analog and Digital Signal Processing, vol. 40, no. 9, pp. 576-579, Sept. 1993.
[47] C. H. Lee and L. H. Chen, “Fast closest codeword search algorithm for vector quantization,” IEE proc. –Vision, Image and Signal Process, vol. 141, no. 3, pp. 143-148, June 1994.
[48] S. J. Baek, B. K. Jeon, and K.-M. Sung, “A fast encoding algorithm for vector quantization,” IEEE Signal Processing Letters, vol. 4, no. 12, pp. 325-327, Dec. 1997.
[49] R. S. Murray, Mathematical Handbook of Formulas and Tables, New York: McGraw-Hill, 1968.