在工業檢測的場合中，自動化影像辨識 (AOI) 技術是一種常見的 應用，現成的影像處理軟體與設備因為授權費與硬體成本問題導致 成本居高不下，並且在產線的需求發生變化時修改的彈性，且準確 度與分析速度並不滿足產線的要求，這份論文中，我們與瑞士的個 人電腦周邊設備供應商「羅技電子」合作，量身設計了新的 AOI 解 決方案，滿足公司的需求。新的解決方案是基於 OpenCV 開發而成， 利用多台視訊攝影機取代昂貴的工業相機，達到了低成本的要求。 本研究所實作出的產品檢驗流程可以用來檢測羅技電子生產的背光 鍵盤的刻字，以及燈光顏色是否符合出貨標準。

In product inspection and quality control, automatic optical inspection (AOI) is a common application. Ready-made AOI softwares and instru- ments exist, but they come with high deployment cost and little flexibility when the requirements are changed. Moreover, the accuracy requirement or inspection time may not meet the specific need of production site. In this thesis, we cooperate with Logitech, a Swiss provider of personal computer and mobile accessories, to design tailor-made AOI solution to meet the company’s need. Our solution is developed based on OpenCV, and utilises multiple web-cams to replace expensive industrial camera, thus achieving a low cost solution. This solution is used to inspect the lettering quality and the LED color of the backlit keyboards from Logitech, so as to determine if the product meets the shipping standard or not.

􀴡謝ii
􁐜要iii
Abstract iv
1 Introduction 1
1.1 Organisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2 Preliminaries 3
2.1 Color Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2 Thresholding and Binarization . . . . . . . . . . . . . . . . . . . . . . 5
2.3 Blob Detection [1] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.4 Feature Point Detection [2] . . . . . . . . . . . . . . . . . . . . . . . . 7
3 Methods 9
3.1 Lettering Defect Detection . . . . . . . . . . . . . . . . . . . . . . . . 9
3.2 LED Defect Detection . . . . . . . . . . . . . . . . . . . . . . . . . . 11
4 Experiments 16
4.1 Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
4.2 Testing Efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
4.3 Testing Robustness . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.4 Testing in the Presence of LED Defects . . . . . . . . . . . . . . . . . 19
4.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
5 Conclusion and Future Work 23
References 25

[1] S. Mallick, “Blob detection using opencv ( python, c++ ).” https://www. learnopencv.com/blob-detection-using-opencv-python-c/.
[2] S. Mallick, “Image alignment (feature based) using opencv (c++/python).” https://www.learnopencv.com/ image-alignment-feature-based-using-opencv-c-python/.
[3] Itseez, The OpenCV Reference Manual, 3.3.1 ed., Oct 2017.
[4] “Rgb color model.” https://en.wikipedia.org/wiki/RGB_color_model, 2018.
[5] “Grayscale.” https://en.wikipedia.org/wiki/Grayscale, 2018.
[6] “Hsl and hsv.” https://en.wikipedia.org/wiki/HSL_and_HSV, 2018.
[7] T. Lindeberg, “Scale invariant feature transform,” 2012.
[8] H. Bay, T. Tuytelaars, and L. Van Gool, “Surf: Speeded up robust features,” in European conference on computer vision, pp. 404–417, Springer, 2006.
[9] E. Rublee, V. Rabaud, K. Konolige, and G. Bradski, “Orb: An efficient alternative to sift or surf,” in Computer Vision (ICCV), 2011 IEEE international conference on, pp. 2564–2571, IEEE, 2011.
[10] P. F. Alcantarilla and T. Solutions, “Fast explicit diffusion for accelerated features in nonlinear scale spaces,” IEEE Trans. Patt. Anal. Mach. Intell, vol. 34, no. 7, pp. 1281–1298, 2011.
[11] A. Kumar, “Computer-vision-based fabric defect detection: A survey,” IEEE transactions on industrial electronics, vol. 55, no. 1, pp. 348–363, 2008.
[12] M. Muja and D. G. Lowe, “Scalable nearest neighbor algorithms for high dimen- sional data,” Pattern Analysis and Machine Intelligence, IEEE Transactions on, vol. 36, 2014.
[13] H.Cantzler,“Randomsampleconsensus(ransac),”InstituteforPerception,Action and Behaviour, Division of Informatics, University of Edinburgh, 1981.