臺灣博碩士論文加值系統

印刷電路板 (printed circuit board, PCB) 是許多電子產品的基礎。自動光學檢測 (automatic optical inspection, AOI) 是判定產品是否存在線路缺點的重要步驟。受限於 AOI 機台的檢驗原理，一般會產生較高的型 I 錯誤 (將良品誤判為不良品)，因此仍需再利用人工的方式進行複判。此種作業方式不只耗時費工，同時仍會存在將不良品誤判為良品 (型 II 錯誤) 之情形。其中，型 II 錯誤的檢驗結果，更是不允許在客戶端被發現。本研究之目的是提出一個分類系統，作為 AOI 檢驗後之智慧缺點分類系統，以解決現行人力需求及誤判所造成的良率損失。本研究利用卷積神經網路及遷移學習，建立一個智慧型缺點判定機制，取代人工複判，期望能降低型 I 及型 II 錯誤的發生，以提升產品良率。本研究採用 VGG16、VGG19、Inception V3、Xception、ResNet50 等模型作為遷移學習之基礎架構。我們以文獻中之公開資料庫，進行分類模型之績效評估及效果驗證。比較結果顯示，Xception 模型之績效最佳，可以獲得 99.6% 的準確率。此結果說明卷積神經網路和遷移學習可被應用於 PCB 缺點之分類。

Printed circuit boards (PCBs) are the foundation of many electronic products. Automatic optical inspection (AOI) is an important step in determining whether there are any defects on PCB. Limited by the inspection principle of the AOI machine, a high type I error (misjudging a good product as a defective product) will generally occur, so it is still necessary to perform a reassessment. This kind of operation is not only time-consuming and labor-intensive, but also involving risks of misclassifying defective products as good products (i.e., type II error). The purpose of this study is to develop a classification model as a smart defect classification system after AOI inspection to solve the current labor demand and yield loss caused by misjudgment. This study uses convolutional neural network and transfer learning to establish an intelligent defect judgment mechanism to replace manual re-judgment, hoping to reduce the occurrence of type I and type II errors and improve product yield. This study applies VGG16, VGG19, Inception V3, Xception, and ResNet50 models as the basic architecture of transfer learning. We conduct performance evaluation of the classification model based on publicly available datasets. The results show that the Xception model performs the best, achieving 99.6% accuracy. This result suggests that convolutional neural networks and transfer learning can be applied to the classification of PCB defects.

中文摘要 i
英文摘要 ii
致謝 iii
圖目錄 vi
表目錄 viii
第一章 緒論 1
1.1 研究背景與動機 1
1.2 研究目的 3
1.3 論文架構 4
第二章 文獻探討 5
2.1 應用卷積神經網路於 PCB 資料缺陷檢測 5
2.2 應用AOI進行PCB圖像辨識 9
第三章 研究方法 11
3.1 卷積神經網路 11
3.2 VGG 模型 16
3.3 ResNet 模型 17
3.4 Inception V3 模型 19
3.5 Xception 模型 20
3.6 績效指標 21
3.7 遷移學習 23
第四章 績效評估 25
4.1 模型 I-VGG16 25
4.2 模型 Ⅱ-VGG19 27
4.3 模型 Ⅲ-ResNet 50 30
4.4 模型 Ⅳ-Inception V3 32
4.5 模型 Ⅴ-Xception 33
第五章 結論與未來研究 37
參考文獻 38

Anoop, K. P., Sarath, N. S. and Kumar, V. V., 2015, A review of PCB defect detection using image processing, International Journal of Engineering and Innovative Technology, 4(11), 188-192.
Adibhatla, V. A., Shieh, J. S., Abbod, M. F., Chih, H. C., Hsu, C. C. and Cheng, J., 2018, Detecting defects in PCB using deep learning via convolution neural networks, In 2018 13th International Microsystems, Packaging, Assembly and Circuits Technology Conference, Taipei, Taiwan, 202-205.
Bianco, S., Cadene, R., Celona, L. and Napoletano, P., 2018, Benchmark analysis of representative deep neural network architectures, IEEE Access, 6, 64270-64277.
Chollet, F. and others, 2015, Keras, https://github.com./fchollet/keras.
Chollet, F., 2017, Xception: deep learning with depthwise separable convolutions, In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Honlulu, HI, USA, 1251-1258.
Dave, N., Tambade, V., Pandhare, B. and Saurav, S., 2016, PCB defect detection using image processing and embedded system, International Research Journal of Engineering and Technology, 3(5), 1897-1901.
Deng, Y. S., Luo, A. C. and Dai, M. J., 2018, Building an automatic defect verification system using deep neural network for PCB defect classification, In 2018 4th International Conference on Frontiers of Signal Processing, Poitiers, France, 145-149.
George, O. D. A., Silva, L. H. D. S., Júnior, A. A., Fernandes, B. J. and Oliveira, S. C., 2020, Binary and multi-label defect classification of printed circuit board based on transfer learning, In Proceedings of 28th European Symposium on Artificial Neural Networks, Bruges, Belgium, 655-660.
Huang, W. and Wei, P., 2019, A PCB dataset for defects detection and classification, Journal of Latex Class Files, 14(8),1-9.
He, K., Zhang, X., Ren, S.and Sun, J., 2016, Deep residual learning for image recognition. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, NV, USA, 770-778.
He, F., Tang, S., Mehrkanoon, S., Huang, X. and Yang, J., 2020, A real-time PCB defect detector based on supervised and semi-supervised learning, In Proceedings of 28th European Symposium on Artificial Neural Networks, Bruges, Belgium, 527-532.
Khalilian, S., Hallaj, Y., Balouchestani, A., Karshenas, H. and Mohammadi, A., 2020, PCB defect detection using denoising convolutional autoencoders, In Proceedings of 2020 International Conference on Machine Vision and Image Processing, Iran, 1-5.
Malge, P. S. and Nadaf, R. S., 2014, A survey: automated visual PCB inspection algorithm, International Journal of Engineering Research & Technology, 3(1), 223-229.
Munisankar, N., Nagarajar, S. and Narendra Kumar Rao, B., 2019, Defect detection in printed board circuit using image processing, International Journal of Innovative Technology and Exploring Engineering, 9(2), 2278-3075.
Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., Blondel, M., Prettenhofer, P., Weiss, R., Dubourg, V., Vanderplas, J., Passos, A., Cournapeau, D., Brucher, M., Perrot, M. and Duchesnay, E., 2001, Scikit-learn:machine learning in Python, Journal of Machine Learning Research, 12, 2825-2830.
Pan, S. J. and Yang, Q., 2009, A survey on transfer learning, IEEE Transactions on Knowledge and Data Engineering, 22(10), 1345-1359.
Rawat, W. and Wang, Z., 2017, Deep convolutional neural networks for image classification: A comprehensive review, Neural Computation, 29(9), 2352-2449.
Simonyan, K. and Zisserman, A., 2014, Very deep convolutional networks for large-scale image recognition, In International Conference on Learning Representations, San Diego, USA, 1150-1220.
Szegedy, C., Vanhoucke, V., Ioffe, S., Shlens, J. and Wojna, Z. 2016, Rethinking the inception architecture for computer vision. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, San Juan, PR, USA, 2818-2826.
Shiina, T., Iwahori, Y. and Kijsirikul, B., 2018, Defect classification of electronic circuit board using multi-input convolutional neural network, International Journal of Computer and Software Engineering, 3(10), 2456-4451.
Silva, L. H. D. S., George, O. D. A., Fernandes, B. J., Bezerra, B. L., Lima, E. B. and Oliveira, S. C., 2019, Automatic optical inspection for defective PCB detection using transfer learning, In IEEE Latin American Conference on Computational Intelligence, Cartagena, Colombia, 1-6.
Taha, E. M., Emary, E. and Moustafa, K., 2014, Automatic optical inspection for PCB manufacturing: a survey, International Journal of Scientific and Engineering Research, 5(7), 1095-1102.
Takada, Y., Shiina, T., Usami, H., Iwahori, Y. and Bhuyan, M. K., 2017, Defect detection and classification of electronic circuit boards using keypoint extraction and CNN features, In The Ninth International Conferences on Pervasive Patterns and Applications Defect, Athens, Greece, 100, 113-116.
Tang, S., He, F., Huang, X. and Yang, J., 2019, Online PCB defect detector on a new PCB defect dataset, arXiv:1902.06197.
Wei, P., Liu, C., Liu, M., Gao, Y. and Liu, H., 2018, CNN-based reference comparison method for classifying bare PCB defects, The Journal of Engineering, 16, 1528-1533.
Wu, K. L. and Hsu, C. Y., 2020, Defect inspection by convolution neural network and empirical study in printed circuit board manufacturing, International Standard on Quality Management, 10(5).
Zhang, L., Jin, Y., Yang, X., Li, X., Duan, X., Sun, Y. and Liu, H., 2018, Convolutional neural network-based multi-label classification of PCB defects, The Journal of Engineering, 16, 1612-1616.
石明于，湯燦泰，黃茂裕，黃泰惠，簡珮珊，2018，「人工智慧於 PCB 瑕疵覆判之應用」，電工通訊季刊，81-95。
陳冠羽，2018，「建構以卷積神經網路為基礎之瑕疵檢測分類模型與實證研究」，桃園，元智大學資訊管理研究所碩士論文。
賴威豪，曾紹崟，2017，「基於深度卷積神經網路之印刷電路板影像瑕疵分類」，2017 AOI論壇與展覽。