本研究的目的在發展太陽能板之印刷線、邊緣及表面瑕疵檢測技術。利用一線型CCD具快速、高解析度及檢測範圍大等特性，結合影像處理演算法，進行瑕疵檢測。在印刷線檢測方面，利用本論文提出的新式Haar-like矩形特徵對太陽能板灰階影像中的細線與粗線做線路檢測。在邊緣瑕疵檢測方面，利用Otsu自動化門檻值做影像分割，再利用B-splin轉折點偵測進行邊緣破裂偵測。在表面瑕疵檢測方面，利用Otsu與valley-emphasis自動化門檻值做三值化圖形分割，經過消除直線後，再以一種以密度為基礎的改良型快速分群演算法(RDBSCAN)及旋轉標記功能做表面瑕疵檢測與分類。本研究以線型CCD所擷取的4096*4096與掃描器所擷取的4096*4096解析度之八角型多晶矽太陽能板影像做驗證。檢測項目包括斷線、針孔、內凹、外凸、刮痕、裂痕、粉塵或汙點與邊緣破裂。經實驗結果得知，線型CCD影像平均每15.7秒可檢測完成，正確率約為85%，而掃描器影像平均每15.1秒可完成檢測，正確率約為90%。

This study focuses on developing an automatic defect detection system on the printed lines, surface and edge of a solar cell. By using a line scan CCD with high speed, high resolution, and vast inspecting scope, a detection system is built and combined with multi-image-processing method to detect the defects. In the printed lines defect detection, this study proposes some new Haar-like features to detect the defects alone the fingers and busbars in a gray-level image. In the edge defect detection, the Otsu automatic thresholding technique is used to segment the image and then the B-spline corner detection is used to detect the edge cracks. In the surface defect detection, Otsu and valley-emphasis automatic thresholding is combined into a three-level thresholding for an input image. After erasing the busbars and fingers in an image, the rapid density-based clustering algorithm (RDBSCAN) with rotated frameworks is proposed to smartly label and sort the surface defects. The proposed methods are evaluated with an octagonal poly-silicon solar cell images (4096*4096) grabbed by a line scan CCD and a scanner as well. The detected items include interruption, pinhole, concave, convex, scratch, fracture, dust or spot, and crack. According to the experimental results, this system can inspect the solar cell image grabbed by a line scan CCD with 85% accuracy rate within 15.7 seconds and that grabbed by a scanner with 90% accuracy rate within 15.1 seconds.

[1] IEA, http://www.iea-pvps.org , 2007.
[2] TEEMA, http://www.teema.org.tw, 2007.
[3] 陳婉如，「台灣太陽電池產業未來展」，光連雙月刊，2005。
[4] 楊昌中，「能源領域中的奈米科技研究」，工研院能源與環境研究所，2005。
[5] Simon Roberts, Solar electricity: A practical guide to designing and installing small photovoltaic systems, PrenticeHall, 1991.
[6] PIDA, http://www.pida.org.tw/, 2007.
[7] Bureau of Energy, Ministry of Economic Affairs, http://www.moeaboe.gov.tw, 2007.
[8] Techlife, http://www.techlife.com.tw, 2007.
[9] 連水養，「量產型單晶太陽電池製作」，國立雲林科技大學電機工程系碩士論文，2002。
[10] 尤如瑾，「世界太陽光電產業現況與展望」，機械工業技術與市場資訊專輯，2005。
[11] 中華太陽能聯誼會，http://www.solar-i.com/know.html， 2007。
[12] 李東獄，「獨立型太陽光電能發電系統分析與研究」，中原大學電機工程學系碩士論文，2007。
[13] 郭冠志，「機器視覺應用於太陽能電池之表面瑕疵檢測」，國立台北科技大學機電整合所碩士論文，2007。
[14] 陳欣怡，「太陽能電池板表面瑕疵檢測」，國立中央大學資訊工程所碩士論文，2007。
[15] 蔡欣儒，「太陽能電池板的尺寸量測與線路瑕疵檢測」，國立中央大學資訊工程研究所，2007。
[16] Z. Fu, Y. Zhao, Y. Liu, Q. Cao, M. Chen, J. Zhang, and J. Lee, "Solar cell crack inspection by image processing," in Proc. Int. IEEE Conf. Business of Electronic Product Reliability and Liability, Shanghai, China, Apr.27-30, 2004, pp.77-80.
[17] N. Otsu, "A threshold selection method from gray-level histograms," IEEE Trans. Systems Man and Cybernetics, vol.9, no.1, pp.62-66, 1979.
[18] P.S. Liao, P.C. Chung, “A fast algorithm for multilevel thresholding.” Journal of Information Science and Engineering, vol 17, No.5, pp. 713-727, 2006.
[19] H.F. Ng “Automatic thresholding for defect detection.”2006 Elsevier, Journal of Pattern Recognition Letters, vol 27,pp. 1649,2006.
[20] P.V.C. Hough, ”Method and Means for Recognizing Complex Patterns,” U.S. Patent 3,069,654, Dec. 18, 1962.
[21] R.O. Duda and P. E. Hart “Use of the HoughTransformation to Detect Lines and Curves in Pictures,”Commun. ACM, 15(1): 11-15, 1972.
[22] T.C. Chen and K.L. Chung.” A New Randomized Algorithm for Detecting Lines,” Real-Time Imaging, 7(6), 2001, pp.473-482.
[23] H.T Sheu and H.Z. Yang, “Open curve segmentation via a two phase scheme”, Pattern Recognition, vol.26, no.12, pp.1839-1844, 1993.
[24] P. Viola, and J,Jones, “Rapid Object Detection Using a Boosted Cascade of Sample Features” IEEE CVPR,Vol. 1, NO, 2, pp 511-518, Dec, 2001.
[25] L. Rainer, and M. Jochen, “An Extended Set of Haar-like Features for Rapid Object Detection” IEEE ICIP 2002, Vol. 1, pp. 900-903, 2002.
[26] A.L.C. Barczak, M.J. Johnson, C.H. Messom, “Real-time Computation of Haar-like features at generic angles for detection algorithms”, Research Letters in the Information and Mathematical Sciences, v. 9, pp. 98 - 111, 2006.
[27] M. Ester, H.P. Kriegel, J. Sander, X. Xu, “A density-based algorithm for discovering clusters in large spatial databases with noise, “ Proceedings of the Second International Conference on Knowledge Discovery and Data Mining, pp. 226–231, 1999.
[28] B. Borah, D.K. Bhattacharyya, “An Improved Sampling-Based DBSCAN for Large Spatial Databases, Proceedings of International Conference on Intelligent Sensing and Information Processing, PP.92 – 96, 2004.
[29] P. Viswanath, R. Pinkesh, “l-DBSCAN : A Fast Hybrid Density Based Clustering Method, “ Pattern Recognition, 2006. ICPR 2006. 18th International Conference on Vol. 1, PP.12 – 915, 2006.
[30] B. Liu, “A Fast Density-Based Clustering Algorithm for Large Databases, “ Machine Learning and Cybernetics, 2006 International Conference, PP.996 – 1000, Aug. 2006.
[31] C.Y. Wu, “An improved Rapid Density-Based Clustering Algorithm and its Application on Real-Time Polarizer Defect Region Defect,” Master Thesis, Graduate Institute of Automation and Control, National Taiwan University of Science and Technology, Taipei, Taiwan, June, 2007.
[32] 鐘國亮，”影像處理與電腦視覺”，東華書局，2004。
[33] 游舜豪，”應用高解析度線掃描CCD於ITO導電玻璃表面瑕疵檢測之研究”，台灣科技大學自動化及控制研究所碩士論文，2006。