近幾年來，由於科技的進步以及半導體製程的演進，晶圓的尺寸已經從原來的6吋演進到目前的12吋。因此整個FOUP (Front Opening Unified Pod)的體積及重量，已經是一般人無法負荷。在目前一般的半導體工廠中，全廠自動化搬運已經是目前的潮流及趨勢。自動化搬運不但可以減少半導體工廠內人力的負擔，並且還可以降低半導體工廠內因人員的走動所產生的灰塵。因此在未來的半導體工廠的設計上，將會朝向全廠搬運自動化的目標。但是在自動化的工廠內，若是有機台發生異常時，由於在自動化的工廠內的人員較少，因此當有人發現異常時，已經常常是數十分鐘之後。如此將會造成工廠產能的損失。本論文的主要目的是在工廠自動化搬運系統中，利用影像識別技術，來提早發現FOUP的異常，減少工廠內製程機台的閒置。本論文所提出的影像辨識系統是利用影像前處理的技術，例如運用動量保持法(moment preserving)來選擇影像灰階之臨界值(Threshold)以去除背景訊息，並運用影像的縮放技術來提高影像中字元的辨識率。在辨識部份,我們利用Zernike Moment辨識區域形狀(region-shape)的效能來識別英文字母及數字字元，並以簡單的Code 39來做軟體的條碼辨識。在實驗部分，我們放置5個等距離的Tag在一台PTZ攝影機前來模擬攝影機辨識置放於不同角度之處理機台上的FOUP之標籤。實驗結果顯示辨識的正確率可達95%以上，因此可以實際應用於自動化半導體工廠的處理機台之FOUP辨識。本系統主要的優點是在擷取影像和辨識字元上具備自動調適的特性，有助於未來實際應用擴充性與適應性。

With the advance of semiconductor manufacture technology, the wafer size has been enlarged from five to twelve inches. In addition, the size and weight of front opening unified pod (FOUP) are also enlarged such that it is burdensome for people to move. This trend to transport enlarged FOUPs makes automatic control a desperate requirement for semiconductor manufacture factories. By constructing automatic FOUP transportation system, the human workload can be reduced largely and the amount of indoor dusts can also be reduced. It’s expected that automatic FOUP transportation would be the basic construction facility for semiconductor companies in the near future. To improve the efficiency of this automation workflow, we proposed to utilize image processing technologies to detect abnormal FOUP carriage to reduce the machine idle time. For image segmentation, it utilized moment preserving algorithm to estimate a proper threshold to remove image backgrounds. It also applies image normalization processes to improve the accuracy of character recognition and identification. For identification, shape-based Zernike moments are used for recognizing English characters and Arabic numerals, and Code 39 is used for barcode identification. For environment setup, five tags place on five boxes are placed in front of one PTZ camera at different angles to simulate the FOUP processing machine. Experiments demonstrate that the identification accuracy is as high as 95%, which shows that the system can work well in semiconductor manufacture factories. The most distinguished feature of the proposed system is its adaptive capability in image acquisition and character recognition, which makes it feasible for practical applications.

[1] 陳傳生，立體物件快速對焦技術，第十九屆機械工程研討會論文集, 2002.
[2] R. Hartley and A. Zisserman, “Multiple view geometry in computer vision,” Cambridge University Press, New York, 2000.
[3] S. Paschalakis and P. Lee, “Pattern recognition in grey level images using moment based invariant features,” Image Processing and Its Applications, vol.1, pp. 245 - 249, July, 1999.
[4] M. R. Teague, “Image analysis via the general theory of moments,” J. Optical Soc. Am., vol.70, pp.920-930, August 1980.
[5] W. H. Tsai, “Moment-preserving threshold:a new approach,” Computer Vision, Graphics and Image Processing, vol.29, no.16, pp. 377, 1985.
[6] Shitu Luo, Qi Zhang, Feilu Luo, Yanling Wang and Zhiyong Chen, “An improved moment-preserving auto threshold image segmentation algorithm,” Proceedings of International Conference on Information Acquisition, pp. 316-318, 2004.
[7] 傅國城，以商標檢索機制實現印鑑之檢索，國立台灣科技大學，電機工程研究所碩士論文，2005.
[8] Dengsheng Zhang Guojun Lu,“Improving retrieval performance of Zernike moment descriptor on affined shapes,” IEEE Int. Conf. Multimedia & Expo. vol.1, pp. 205- 208, 2002.
[9] S. X. Liao and M. Pawlak, “Image analysis with Zernike moment descriptors,” IEEE Canadian Conference on Electrical and Computer Engineering, vol. 2, pp. 700-703, 25th–28th May 1997.
[10] S. X. Liao and M. Pawlak, “On the accuracy of Zernike moments for image Analysis,” IEEE Trans Pattern Analysis and Machine Intelligence, vol. 20, no. 12, December 1998.
[11] W. Kim and Y. Kim, “A region-based shape descriptor using Zernike moments,” Signal Processing: Image Communication 16, pp 95-102, 2000.
[12] R. Muniz, L. Junco and A. Otero, “A robust software barcode reader using the Hough transform,” Proceedings of 1999 International Conference, 31 Oct.-3 pp. 313-319, Nov. 1999.
[13] X. Fang, F. Wu, B. Luo, H. Zhao and P. Wang, “Automatic recognition of noisy Code-39 barcode,” International Conference on Artificial Reality and Telexistence, , pp. 79-82, Nov. 2006.
[14] 劉鴻明，蔡孟達，張元翔，應用於影像縮放技術之內插法評估研究，中原大學資訊工程學系，2005.
[15] 張銘豪，利用分割辨識方法之英文數字系統，國立中山大學，資訊工程研究所碩士論文，1996.
[16] C. Coetzee, C. Botha and D. Weber, “PC based number plate recognition system,” in Proc. IEEE Symposium on Industrial Electronics, vol.2, pp. 605-610, 1998.