隨著自動化發展的趨勢，人工目視檢測近年來逐漸被自動光學檢測取代，以提升生產速度並降低組裝成本。然而現階段設備普遍成本高昂，對於少量檢測需求的研發實驗室來說實在難以負擔。除此之外，基於配合度與成本考量，相較於通用式的學習型檢測系統，比較特殊或比較精密的物體通常會採用客製化設計的系統。
　　因此本論文針對高成本且精密度高之表面黏著型(SMD)板端連接器I-PEX 20474-030E進行檢測系統開發之研究，檢測項目包含，缺件(Component Missing)、歪斜(Skewing)、極性反(Wrong Priority)、偏移(Component Shifting)、橋接(Solder Bridge)與空焊(Missing Solder)。以離線機台應用為目標設計，操作方式簡單化，無需人工事先定位，並致力於降低硬體設備成本。檢測技術部分整合數學演算法與影像處理技術，以快速連通標記法取得檢測物，經由修正過的最小方差法來計算擺放的旋轉角度並自動轉正檢測物，最後使用形態學等影像處理技術完成圖像的分析判斷並檢測出連接器的瑕疵。
　　本研究之實驗對1920張樣本影像進行檢測，其結果顯示系統檢測率可高達96.57%，誤判率為1.13%，並可抵抗雜訊以及光源亮度變化，是兼具準確性、穩定性與操作便利性的高速檢測系統。

　With the trend of automation, manual vision inspection (MVI) has been gradually replaced by automatic optical inspection (AOI) for speeding up the production and reducing the cost of assembly nowadays. However, equipments are usually expensive, which will be a heavy burden for a small- amount demand R&D laboratory in a company. As for current choice of AOI equipments, customized ones are often more welcome due to the consideration of fitness and cost, rather than a universal learning system, especially when the objects to be tested are special or high precision.
　Our study focuses on developing an inspection system for a surface-mounted device receptacle connector, I-PEX 20474-030E, which is a high-cost and high-precision electronic component. The defects to be detected include component missing, skewing, wrong priority, component shifting, solder bridge, and missing solder. The proposed system is designed as an off-line inspection. Manual positioning is not required before detection to make the operation easy and simple and the total costs of equipments are kept low in our system. For the algorithms of the system, it is combined by mathematical algorithms and image processing technique. First the target object is obtained by fast connected component, and then the revised least square error method calculate the inclination angles of the object to help positioning the freely-placed connectors. After that, morphology and other image processing technique are applied for image analysis to detect the defects of the connectors.
　With 1920 images of test samples, the experimental results indicate that the detection rate of the proposed method for each defect is at 96.57%, and the overall false rate is 1.13%. The method has little interference of noise or changes in brightness, hence it can be said to be a reliable and stable real-time inspection system with simple and convenient operation.

[1]非凡傳播公司，"產業科技通 - AOI檢測"，2003。
[2]林定皓，"印刷電路板電子組裝技術概述"，台灣電路板協會，2006。
[3]林定皓，"電路板電氣測試與AOI檢驗技術簡介"，台灣電路板協會，2008。
[4]林姿君，"自動化光學檢測專業能力指標與課程發展之硏究"，台灣科技大學電機工程系碩士論文，2009。
[5]M. Moganti, F. Ercal, C. H. Dagli, and S. Tsunekawa, "Automatic PCB Inspection Algorithms: A Survey," Computer Vision and image Understanding, vol. 63, no. 2, pp. 287-313, 1996.
[6]R. T. Chin, "Automated Visual Inspection: A Survey," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. PAMI-4, no.6, pp. 557-573, 1982.
[7]R. T. Chin, "Survey: Automated Visual Inspection: 1981 to 1987," Computer Vision, Graphics, and Image Process, vol. 41, no.3, pp. 346-381, 1988.
[8]T. H. Kim. T. H. Cho, Y. S. Mon, and S. H. Park, "Visual Inspection System for The Classification of Solder Joints," Pattern Recognition, vol. 32, no.4, pp. 565-575, 1999.
[9]W. Y. Wu, M. J. Wang, and C. M. Liu, "Automated Inspection of Printed Circuit Boards through Machine Vision," Computers in Industry, vol. 28, no. 2, pp. 103-111, 1996.
[10]黃志榮，"自動檢測之定位、量測與辨識之研究"，國立成功大學電機工程學系碩士論文，2003。
[11]S. L. Bartlett, P. J. Besl, C. L. Cole, R. Jain, D. Mukherjee, and K. D. Skifstad, "Automatic Solder Joint Inspection," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 10, no. 1, pp. 31-43, 1988.
[12]許光城、林錦彣、陳建元，"牙線棒瑕疵自動光學檢測之研究"，第十一屆全國AOI論壇與展覽大會手冊，2012。
[13]李承祐，"表面黏著元件自動化檢測"，義守大學工業工程與管理學系碩士論文，2006。
[14]劉嘉唐，"BGA錫球定位與幾何形狀之視覺檢測方法研究"，元智大學機械工程學系碩士論文，2005。
[15]張振堶、金美敬，自動光學檢測(AOI)市場及技術發展趨勢調查，工研院IEK系統能源組，2002。
[16]S. K. Nayar, M. Gupta, "Diffuse Structured Light," IEEE International Conference on Computational Photography, pp. 1-8, 2012.
[17]照明節能產品應用手冊，財團法人台灣綠色生產力基金會，http://www.ecct.org.tw/print/index.htm，2011。
[18]Machine Vision Optics Data 5, VS Technology Corp., 2009.
[19]CCS Inc., http://www.ccs-inc.co.jp/, Jun 25, 2013.
[20]鍾國亮，影像處理與電腦視覺，4th ed，台灣東華書局股份有限公司，2011。
[21]R. C. Gonzalez, R. E. Woods, Digital Image Processing, 3rd ed. New Jersey: Prentice Hall, 2008.
[22]N. Otsu, "A Threshold Selection Method from Gray Level Histogram," IEEE Transactions on System, Man and Cybernetics, vol. 9, pp. 62-67, 1979.
[23]廖柏宇，"即時智慧型手勢提取與辨識系統"，台灣科技大學電機工程系碩士論文，2011。
[24]L. He, Y. Chao, and K. Suzuki, "A Run-Based Two-Scan Labeling Algorithm," IEEE Transactions on Image Processing, vol. 17, no. 5, pp. 749-756, 2008.
[25]L. He, Y. Chao, K. Suzuki, K. Wu, "Fast Connected-Component Labeling," Pattern Recognition, vol. 42, no. 9, pp. 1977-1987, 2009.
[26]Machine Vision Optics Data Asia.1, VS Technology Corp., 2011.
[27]Dai-ichi Seiko Co., Ltd, http://www.i-pex.com/, Jun 30, 2013.