在嚴苛的半導體製程條件下，誤將錯誤的鑽石修整器裝上機台將造成不可設想的後果，為避免將錯誤的規格批號出給客戶，本研究探討光學字符辨識（OCR）系統在金屬表面雷射標籤辨識中的應用，以及使用物件偵測YOLOv7(You Only Look Once)模型作為替代方法的可行性。通過改進標籤資料的處理方法和引入資料增強技術，旨在克服環境變因對辨識準確率的影響，從而提高金屬表面標籤辨識的效能。
在本研究中，優化了標籤資料的處理流程，利用OCR系統辨識文字位置和預測文字，並透過等距分割方法獲得每個字符的獨立座標點外框，以轉換為YOLOv7模型可識別的格式。同時，針對圖片的影像預處理採用了圖像畫像素調整和圖像切割處理，以減低圖像雜訊對模型的干擾，並減少非辨識物件對圖像複雜度的影響。此外，通過調整YOLOv7資料增量超參數並隨機生成訓練圖片，以應對多種環境變因對辨識率的影響。
實驗結果顯示，本研究基於YOLO物件偵測模型提出的優化標籤資料方法在金屬表面標籤辨識中取得了相當高的辨識水準。YOLOv7模型的準確率(Precision)、召回率(Recall)和平均精度均值(mean Average Precision)分別達到了0.83、0.86和0.99。與其他OCR方法相比，YOLOv7模型在規格名稱和Lot-ID的平均單詞正確率表現更為優秀。
總體而言，本研究所提出的資料標籤改良方法結合資料增強技術，對金屬表面雷射標籤辨識具有重要意義，研究成果為未來金屬表面辨識相關研究提供了有價值的參考。

Under stringent semiconductor manufacturing conditions, incorrect installation of the wrong diamond dresser on the machine can have unforeseeable consequences. To avoid issuing incorrect specification batch numbers to customers, exercising utmost caution is crucial.
This study explores Optical Character Recognition (OCR) for laser-engraved metal labels, along with the feasibility of using the YOLOv7 object detection model. We optimized the label data processing workflow, identifying text positions and characters with OCR, and applying equidistant segmentation for independent character bounding boxes. Image preprocessing involved resizing and segmentation to reduce noise interference. Adjusting YOLO data augmentation hyperparameters and generating training images addressed recognition rate influences from environmental factors.
Experimental results demonstrate that the proposed approach achieves a notably high level of recognition accuracy in metal surface label recognition. The precision, recall, and mean Average Precision (mAP) of the YOLOv7 model reached 0.83, 0.86, and 0.99, respectively. Comparative analysis with other OCR methods indicates superior performance of the YOLOv7 model in terms of average word accuracy for specification names and Lot-IDs. Overall, the combination of data label refinement and augmentation techniques presented in this study holds significant implications for metal surface laser label recognition.

金屬表面雷射標籤辨識基於YOLO物件偵測方法之研究 i
摘要 i
誌謝 iii
目錄 iv
表目錄 v
圖目錄 vi
第一章 緒論 1
1.1 研究背景 1
1.2 研究動機 3
1.3 研究目的 5
1.4 論文架構 6
第二章 文獻回顧 8
2.1 金屬工件圖像獲取 8
2.2 金屬表面圖像預處理 10
2.3 光學字符辨識OCR 12
2.4 金屬工件表面OCR 15
2.5 YOLO物件偵測模型 16
2.6 物件偵測方法應用於文字識別 25
2.7 文獻回顧小結 26
第三章 問題描述 27
3.1 圖像獲取設備限制 27
3.2 圖像雜訊問題 29
3.3 OCR辨識雷射刻字資料面臨問題 31
第四章 研究方法 33
4.1 資料蒐集與圖像預處理 33
4.2 光學字符辨識(OCR) 35
4.3 文字識別座標轉換 36
4.4 資料增強 38
4.5 評估指標 40
第五章 實驗結果與分析 43
5.1 資料集說明 43
5.2 實驗環境 43
5.3 YOLOv7參數設定 44
5.4 YOLOv7訓練結果 46
5.5 YOLOv7與OCR辨識結果比較 48
第六章 結論與未來展望 53
6.1 結論 53
6.2 後續研究方向 55
參考文獻 56
附錄一 規格名稱辨識結果 63
附錄二 Lot-ID 辨識結果 66

[1] Zhong, Zhao-Wei. "Recent developments and applications of chemical mechanical polishing." The International Journal of Advanced Manufacturing Technology 109.5 (2020): 1419-1430.
[2] Wang, Wantang, et al. "Improvement in dispersion stability of alumina suspensions and corresponding chemical mechanical polishing performance." Applied Surface Science 597 (2022): 153703.
[3] Shi, Xiaolei, et al. "Characterization of colloidal silica abrasives with different sizes and their chemical–mechanical polishing performance on 4H-SiC (0 0 0 1)." Applied Surface Science 307 (2014): 414-427.
[4] Chen, Shidong, and Hong Lei. "Polishing performance and mechanism of a novel Fe-based slurry for chemical mechanical polishing." Tribology International 194 (2024): 109549.
[5] Holesovsky, Frantisek, et al. "Evaluation of diamond dressing effect on workpiece surface roughness by way of analysis of variance." Tehnički vjesnik 25.Supplement 1 (2018): 165-169.
[6] Tao, Xian, et al. "Automatic metallic surface defect detection and recognition with convolutional neural networks." Applied Sciences 8.9 (2018): 1575.
[7] Lv, Xiaoming, et al. "Deep metallic surface defect detection: The new benchmark and detection network." Sensors 20.6 (2020): 1562.
[8] Xu, Yiming, Kai Zhang, and Li Wang. "Metal surface defect detection using modified YOLO." Algorithms 14.9 (2021): 257.
[9] Zhong, Zhao-Wei. "Recent developments and applications of chemical mechanical polishing." The International Journal of Advanced Manufacturing Technology 109.5 (2020): 1419-1430.
[10] Liang, H., et al. "Wear phenomena in chemical mechanical polishing." Wear 211.2 (1997): 271-279.
[11] Boncelet, Charles. "Image noise models." The essential guide to image processing. Academic Press, 2009. 143-167.
[12] Verma, Rohit, and Jahid Ali. "A comparative study of various types of image noise and efficient noise removal techniques." International Journal of advanced research in computer science and software engineering 3.10 (2013).
[13] Xiang, Zhong, Huaxiong Wu, and Ding Zhou. "Recognition of characters on curved metal workpiece surfaces based on multi-exposure image fusion and deep neural networks." Neurocomputing 512 (2022): 117-129.
[14] Senin, Nicola, Adam Thompson, and Richard K. Leach. "Characterisation of the topography of metal additive surface features with different measurement technologies." Measurement Science and Technology 28.9 (2017): 095003.
[15] Xu, Yiming, Kai Zhang, and Li Wang. "Metal surface defect detection using modified YOLO." Algorithms 14.9 (2021): 257.
[16] Fang, Xiaoxin, et al. "Research progress of automated visual surface defect detection for industrial metal planar materials." Sensors 20.18 (2020): 5136.
[17] Piironen, T.; Silven, O.; Pietikainen, M.; Laitinen, T.; Strommer, E. Automated visual inspection of rolled metal surfaces. Mach. Vis. Appl. 1990, 3, 247–254. [CrossRef]
[18] Xu, Qiong., Chen, Jing., Zheng, Boyu. "Metal can image acquisition device and acquisition method." null (2016).
[19] Zhang, X.; Ding, Y.; Lv, Y.; Shi, A.; Liang, R. A vision inspection system for the surface defects of strongly reflected metal based on multi-class SVM. Expert Syst. Appl. 2011, 38, 5930–5939.
[20] Luo, Qiwu, and Yigang He. "A cost-effective and automatic surface defect inspection system for hot-rolled flat steel." Robotics and Computer-Integrated Manufacturing 38 (2016): 16-30.
[21] Xiang, Zhong, et al. "Metal stamping character recognition algorithm based on multi-directional illumination image fusion enhancement technology." EURASIP Journal on Image and Video Processing 2018 (2018): 1-11.
[22] Sonka, Milan, et al. "Image pre-processing." Image processing, analysis and machine vision (1993): 56-111.
[23] Guirong, Weng., Ping, Ye. "Image processing and classification of metal materials based on mathematical morphology." (2008). doi: 10.1117/12.819709
[24] Nomura, Yukimasa., Kishimoto, Kyoji. "Method for image processing inspection of glossy metal surface." (1995).
[25] Lingliang, Mei., Dihua, Xu., Boping, Chen., Tianhui, Wang., Wei, Wang. "Method for processing metallic phase image information." (2009).
[26] Hagiwara, Yoshichi., Kimura, Naoya., Kenji, Emori. "Process for formation of metallic relief." (1987).
[27] Nagatomo, Yoshihiro. "Processing method for surface of metal." (1996).
[28] Gross, Ralph, and Vladimir Brajovic. "An image preprocessing algorithm for illumination invariant face recognition." International Conference on Audio-and Video-Based Biometric Person Authentication. Berlin, Heidelberg: Springer Berlin Heidelberg, 2003.
[29] Grönlund, Jakob, and Angelina Johansson. "Defect Detection and OCR on Steel." (2019).
[30] Yu, Jiahui, et al. "Automatic and efficient metallic surface defect detection based on key pixel point locations." IEEE Sensors Journal 21.10 (2020): 11476-11487.
[31] Nguyen, T. T. H., Jatowt, A., Coustaty, M., & Doucet, A. (2021). Survey of post-OCR processing approaches.ACM Computing Surveys (CSUR), 54(6), 1-37.
[32] Patel, C., Patel, A., & Patel, D. (2012). Optical character recognition by open source OCR tool tesseract: A case study.International Journal of Computer Applications, 55(10), 50-56.
[33] Krishnamoorthi, M., et al. "Improving Optical Character Recognition (OCR) Accuracy using Multi-Layer Perceptron (MLP)." 2023 7th International Conference on Trends in Electronics and Informatics (ICOEI). IEEE, 2023.
[34] Shukla, Ananya, et al. "Devnagari Character Recognition using Optical Character Recognition (OCR)." 2023 International Conference on Computational Intelligence, Communication Technology and Networking (CICTN). IEEE, 2023.
[35] Patel, Chirag, Atul Patel, and Dharmendra Patel. "Optical character recognition by open source OCR tool tesseract: A case study." International Journal of Computer Applications 55.10 (2012): 50-56.
[36] Mendel, Gili. Optical character recognition using morphological attributes. Louisiana State University and Agricultural & Mechanical College, 1993.
[37] Robby, G. Abdul, et al. "Implementation of optical character recognition using tesseract with the javanese script target in android application." Procedia Computer Science 157 (2019): 499-505.
[38] Dome, Saurabh, and Asha P. Sathe. "Optical charater recognition using tesseract and classification." 2021 International Conference on Emerging Smart Computing and Informatics (ESCI). IEEE, 2021.
[39] Poncelas, Alberto, et al. "A tool for facilitating ocr postediting in historical documents." arXiv preprint arXiv:2004.11471 (2020).
[40] Kamisetty, Venkata Naga Sai Rakesh, et al. "Digitization of Data from Invoice using OCR." 2022 6th International Conference on Computing Methodologies and Communication (ICCMC). IEEE, 2022.
[41] Mithe, Ravina, Supriya Indalkar, and Nilam Divekar. "Optical character recognition." International journal of recent technology and engineering (IJRTE) 2.1 (2013): 72-75.
[42] Kuang, Z., Sun, H., Li, Z., Yue, X., Lin, T. H., Chen, J., ... & Lin, D. (2021, October). MMOCR: a comprehensive toolbox for text detection, recognition and understanding. In Proceedings of the 29th ACM International Conference on Multimedia (pp. 3791-3794).
[43] Francisco, Alexandre, Umberto Biccari, and Enrique Zuazua. "Computational performance of the MMOC in the inverse design of the Doswell frontogenesis equation." arXiv preprint arXiv:2210.03798 (2022).
[44] Kumari, Preeti, Nitin Jonathan Myers, and Robert W. Heath. "Adaptive and fast combined waveform-beamforming design for mmWave automotive joint communication-radar." IEEE Journal of Selected Topics in Signal Processing 15.4 (2021): 996-1012.
[45] Aldrich, S., Arowolo, M. O., He, F., Popescu, M., & Xu, D. (2022, December). Comprehensive Assessment of OCR Tools for Gene Name Recognition in Biological Pathway Figures. In 2022 IEEE International Conference on Bioinformatics and Biomedicine (BIBM) (pp. 3574-3579). IEEE.
[46] Du, Y., Li, C., Guo, R., Yin, X., Liu, W., Zhou, J., ... & Wang, H. (2020). Pp-ocr: A practical ultra lightweight ocr system. arXiv preprint arXiv:2009.09941.
[47] Du, Y., Li, C., Guo, R., Cui, C., Liu, W., Zhou, J., ... & Ma, Y. (2021). Pp-ocrv2: Bag of tricks for ultra lightweight ocr system. arXiv preprint arXiv:2109.03144.
[48] Feng, L., Ke, Z., & Wu, N. (2022, July). ModelsKG: A Design and Research on Knowledge Graph of Multimodal Curriculum Based on PaddleOCR and DeepKE. In 2022 14th International Conference on Advanced Computational Intelligence (ICACI) (pp. 186-192). IEEE.
[49] Guo, Qingze. "Technology for making matel plane decorative exquisite photograph and scripts and paintings." (1999).
[50] Hwang, Young, Kyu., Hong, Do, Young., Kwak, Jae, Sung., Oh, Kyung, Ryul., Chang, Jong, San., Kwon, Yong, Uk., Cho, Kyung, Ho., Lee, U, Hwang., Hwang, Dong, Won., Lee, Su, Kyung., Sanil, E, S, Eriyakkadan, Sivan. "- - - The complex matel-organic trimer/support complex formed through covalant bonding between support and organic-metal trimer." (2019).
[51] Li, Jing, et al. "Detection and recognition of characters on the surface of metal workpieces with complex background." 2020 IEEE 4th information technology, networking, electronic and automation control conference (ITNEC). Vol. 1. IEEE, 2020.
[52] Beyazian, Jeanne, and John Sadi. "OCR for laser marking quality assessment." Sixteenth International Conference on Quality Control by Artificial Vision. Vol. 12749. SPIE, 2023.
[53] Luo, Hongliang, et al. "YOLO: An efficient terahertz band integrated sensing and communications scheme with beam squint." IEEE Transactions on Wireless Communications (2024).
[54] Li, Z., Yuan, J., Li, G., Wang, H., Li, X., Li, D., & Wang, X. (2023). RSI-YOLO: object detection method for remote sensing images based on improved YOLO. Sensors, 23(14), 6414.
[55] Wang, Xiuling, et al. "ACP-YOLO: Asymmetric Center Point Bounding Box Regression Strategy and Angle Loss-based YOLO for Object Detection." (2023).
[56] Wang, Z., Zhang, X., Li, J., & Luan, K. (2021). A YOLO-based target detection model for offshore unmanned aerial vehicle data. Sustainability, 13(23), 12980.
[57] Ponika, Mannem, et al. "Developing a YOLO based Object Detection Application using OpenCV." 2023 7th International Conference on Computing Methodologies and Communication (ICCMC). IEEE, 2023.
[58] K., Venkateswarlu. "YOLO Based Advanced Smart Traffic Assistance Platform for Number Plate and Helmet Detection." International Journal For Science Technology And Engineering, (2023).
[59] Maji, D., Nagori, S., Mathew, M., & Poddar, D. (2022). Yolo-pose: Enhancing yolo for multi person pose estimation using object keypoint similarity loss. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 2637-2646).
[60] Redmon, J., Divvala, S., Girshick, R., & Farhadi, A. (2016). You only look once: Unified, real-time object detection. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 779-788).
[61] Redmon, Joseph, et al. "You only look once: Unified, real-time object detection." Proceedings of the IEEE conference on computer vision and pattern recognition. 2016.
[62] Redmon, Joseph, and Ali Farhadi. "YOLO9000: better, faster, stronger." Proceedings of the IEEE conference on computer vision and pattern recognition. 2017.
[63] Sang, J., Wu, Z., Guo, P., Hu, H., Xiang, H., Zhang, Q., & Cai, B. (2018). An improved YOLOv2 for vehicle detection. Sensors, 18(12), 4272..
[64] Huang, X., Wang, X., Lv, W., Bai, X., Long, X., Deng, K., ... & Yoshie, O. (2021). PP-YOLOv2: A practical object detector. arXiv preprint arXiv:2104.10419.
[65] Redmon, Joseph, and Ali Farhadi. "Yolov3: An incremental improvement." arXiv preprint arXiv:1804.02767 (2018).
[66] Zhao, Liquan, and Shuaiyang Li. "Object detection algorithm based on improved YOLOv3." Electronics 9.3 (2020): 537.
[67] Mao, Q. C., Sun, H. M., Liu, Y. B., & Jia, R. S. (2019). Mini-YOLOv3: real-time object detector for embedded applications. Ieee Access, 7, 133529-133538.
[68] Li, C., Wang, R., Li, J., & Fei, L. (2020). Face detection based on YOLOv3. In Recent Trends in Intelligent Computing, Communication and Devices: Proceedings of ICCD 2018 (pp. 277-284). Springer Singapore.
[69] Bochkovskiy, Alexey, Chien-Yao Wang, and Hong-Yuan Mark Liao. "Yolov4: Optimal speed and accuracy of object detection." arXiv preprint arXiv:2004.10934 (2020).
[70] Yu, J., & Zhang, W. (2021). Face mask wearing detection algorithm based on improved YOLO-v4. Sensors, 21(9), 3263.
[71] Wu, D., Lv, S., Jiang, M., & Song, H. (2020). Using channel pruning-based YOLO v4 deep learning algorithm for the real-time and accurate detection of apple flowers in natural environments. Computers and Electronics in Agriculture, 178, 105742.
[72] Gai, R., Chen, N., & Yuan, H. (2023). A detection algorithm for cherry fruits based on the improved YOLO-v4 model. Neural Computing and Applications, 35(19), 13895-13906.
[73] Wu, W., Liu, H., Li, L., Long, Y., Wang, X., Wang, Z., ... & Chang, Y. (2021). Application of local fully Convolutional Neural Network combined with YOLO v5 algorithm in small target detection of remote sensing image. PloS one, 16(10), e0259283.
[74] Kim, J. H., Kim, N., Park, Y. W., & Won, C. S. (2022). Object detection and classification based on YOLO-V5 with improved maritime dataset. Journal of Marine Science and Engineering, 10(3), 377.
[75] Wu, T. H., Wang, T. W., & Liu, Y. Q. (2021, June). Real-time vehicle and distance detection based on improved yolo v5 network. In 2021 3rd World Symposium on Artificial Intelligence (WSAI) (pp. 24-28). IEEE.
[76] Li, C., Li, L., Jiang, H., Weng, K., Geng, Y., Li, L., ... & Wei, X. (2022). YOLOv6: A single-stage object detection framework for industrial applications. arXiv preprint arXiv:2209.02976.
[77] Zeng, Yalin, et al. "Yolov7-uav: An unmanned aerial vehicle image object detection algorithm based on improved yolov7." Electronics 12.14 (2023): 3141. [54] Jingyang, Wang., Naiwen, Yu. "A Mask-Wearing Detection Model in Complex Scenarios Based on YOLOv7-CPCSDSA." Electronics, (2023).
[78] Zhu, Qiang, et al. "YOLOv7-CSAW for maritime target detection." Frontiers in neurorobotics 17 (2023).
[79] Sun, S., Ma, H., Wang, K., Wang, C., Wang, Z., & Yuan, H. (2023). Research on the Detection Method of Coal Mine Roadway Bolt Mesh Based on Improved YOLOv7. Electronics, 12(14), 3050.
[80] Kachurova, Monika, et al. "Comparative analysis of YOLOv7 models for use in perimeter protection of military areas." 2023 30th International Conference on Systems, Signals and Image Processing (IWSSIP). IEEE, 2023.
[81] Wang, C. Y., Bochkovskiy, A., & Liao, H. Y. M. (2023). YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (pp. 7464-7475).
[82] Jacob Solawetz. (Jan 4, 2024). What is YOLOv7? A Complete Guide.. Roboflow Blog: https://blog.roboflow.com/yolov7-breakdown/
[83] Laroca, R., Severo, E., Zanlorensi, L. A., Oliveira, L. S., Gonçalves, G. R., Schwartz, W. R., & Menotti, D. (2018, July). A robust real-time automatic license plate recognition based on the YOLO detector. In 2018 international joint conference on neural networks (ijcnn) (pp. 1-10). IEEE.
[84] 陳沛樺（2009）。化學機械拋光中先進鑽石碟對石墨拋光墊之修整與拋光效能研究。﹝碩士論文。國立清華大學﹞
[85] 楊琦婷（2002）。化學機械研磨中鑽石修整器磨耗之研究。﹝碩士論文。國立臺灣大學﹞