臺灣博碩士論文加值系統

科技不斷進步，智慧型產品與人類生活息息相關，如手機、平板電腦與智慧手錶。由於防水、耐用、高透光與不易被腐蝕等特性，使得玻璃成為智慧型產品螢幕表面之保護材料。為了提高產品的品質與良率，檢測成為生產線上不可或缺的重要環節；然而，人工檢測的主觀性判斷使得檢測標準參差不齊，且其帶來的高人事成本，促使自動化檢測導入漸漸成為一種趨勢。
自動化檢測中，以影像處理相關之演算法為現行最被廣泛使用的技術。仰賴長期所累積的知識與經驗，影像處理技術具備解決大部分瑕疵檢測的能力；但是，在一般生產環境下，影像處理技術易受光源與環境影像，使參數設定泛用性降低，且愈來愈嚴格的檢測標準將提高研發成本並拉長研發所需時間。為了克服現行傳統影像處理演算法技術瓶頸，本研究提出一應用深度學習技術之手機玻璃表面瑕疵檢測方法。研究架設包括取像系統、光源與電腦。本研究執行步驟，首先進行影像前處理，包括取像影像分割、資料強化、瑕疵分類等步驟建立影像資料庫；緊接著，進行神經網路訓練以不同的卷積神經網路與優化器訓練獲得最佳手機玻璃表面瑕疵檢測方法。本研究展現如何建立一以深度學習技術檢測手機玻璃表面瑕疵的方法與流程，並透過神經網路訓練結果證實其應用於手機玻璃表面瑕疵的突破。

Inspection systems have been widely applied in industries to examine defects in products. Investigations into the method of defect inspection are important for manufacturing process. In this paper, we present an inspection approach based on deep learning, with a view to distinguishing flawed smartphone cover glass and to classify different kinds of glass defects. In contrast with traditional techniques, image processing techniques (IPTs), which often creates challenges to algorithm calculations due to inconsistent real-world variables like lighting and shadow changes, deep learning-based defect inspection technique is less time-consuming. This study proposes a method using deep learning architecture of convolutional neuron networks (CNNs) for detecting cover glass defects. We cropped the images into pieces sized 300"×" 300 pixel resolutions and classified defects features like scratches, dust, dirt and fur, which were subsequently learned automatically by different models of CNN, including LeNet-5, AlexNet, VGG and GoogLeNet. As the CNN finished auto-learning, it was then capable of detecting and classifying different kinds of defects. The best results among the used models showed that the accuracy of a trained network was 98% when classifying cover glass defects in the right categories, proven to perform better compared with IPT. This paper concludes that the proposed method can indeed find the defects in realistic situations with a more ease manner.

口試委員會審定書 I
致謝 II
摘要 III
Abstract IV
List of Figures VIII
List of Tables X
Chapter 1 Introduction 1
1.1 Background 1
1.1.1 The Trend of Inspection in Electronic Devices 1
1.1.2 The Rise of Artificial Intelligence Technology 3
1.2 Motivation 5
1.3 Objectives 6
1.4 Literature Review 6
1.5 Organization 12
Chapter 2 Convolutional Neural Network 13
2.1 Layer 13
2.1.1 Convolutional Layer 14
2.1.2 Pooling Layer 16
2.1.3 Fully Connected Layer 17
2.1.4 Activation Function 18
2.1.5 Softmax Layer 21
2.2 Architecture 22
2.2.1 LeNet-5 22
2.2.2 AlexNet 23
2.2.3 Visual Geometry Group 24
2.2.4 GoogLeNet 26
2.3 Gradient Descent 29
2.3.1 Stochastic Gradient Descent 31
2.3.2 Gradient Descent with Momentum 32
2.3.3 Root Mean Square Propagation 34
2.3.4 Adaptive Moment Estimation 34
2.3.5 Nesterov Accelerated Gradient 36
Chapter 3 Methods 39
3.1 Equipment 39
3.2 Overview of the Method 42
3.3 Image Preprocessing 44
3.3.1 Image Capture 44
3.3.2 Image Segmentation 45
3.3.3 Image Classification 46
3.3.4 Image Augmentation 47
3.4 Deep Neural Network Training 50
3.4.1 Build Database 50
3.4.2 Neural Network Training 52
Chapter 4 Results and Discussion 57
4.1 Build Database 57
4.1.1. Proper Dataset Split Ratio 57
4.1.2. Desirable Number of Training Images 59
4.2 Training Results 63
Chapter 5 Conclusions and Future Works 71
5.1 Conclusions 71
5.2 Future Works 72
References 73

[1] Statista Research Department, "Worldwide tablet shipments from 2nd quarter 2010 to 3rd quarter 2018," Statista, 7 November 2018. [Online]. Available: https://www.statista.com/. [Accessed February 2018].
[2] Statista Research Department, "Global smartphone shipments from 2009 to 2018," Statista, 1 February 2019. [Online]. Available: https://www.statista.com/. [Accessed February 2019].
[3] Statista Research Department, "Smartwatch unit sales worldwide from 2014 to 2018," Statista, 3 May 2019. [Online]. Available: https://www.statista.com/. [Accessed May 2019].
[4] No Author, "Automated optical metrology market by product, by end-use application - growth, share, opportunities & competitive analysis 2016 - 2023," Credence Research, July 2016. [Online]. Available: https://www.credenceresearch.com/. [Accessed December 2018].
[5] No Author, "Artificial intelligence software market to reach $89.8 billion in annual worldwide revenue by 2025," Tractica, 21 December 2017. [Online]. Available: https://www.tractica.com/. [Accessed December 2018].
[6] No Author, "Artificial intelligence implementations have expanded to encompass 215 discrete use cases across virtually all industries," Tractica, 28 August 2017. [Online]. Available: https://www.tractica.com/. [Accessed December 2018].
[7] P. Xiangqian, Y. Chen, W. Yu, Z. Zude and G. Sun, "An online defects inspection method for float glass fabrication based on machine vision," The International Journal of Advanced Manufacturing Technology, vol. 39, pp. 1180-1189, 2008.
[8] F. Adamo, F. Attivissimo, A. D. Nisio and M. Savino, "A low-cost inspection system for online defects assessment in satin glass," Measurement, vol. 42, pp. 1304-1311, 2009.
[9] J. Chuanxia, G. Jian and A. Yinhui, "Automatic surface defect detection for mobile phone," Applied Soft Computing, vol. 52, pp. 348-358, 2017.
[10] D. Ziou and S. Tabbone, "Edge detection techniques - an overview," Pattern Recognition and Image Analysis C/C of Raspoznavaniye Obrazov I Analiz Izobrazhenii, vol. 8, pp. 537-559, 1998.
[11] Y. LeCun, L. Bottou, Y. Bengio and P. Haffner, "Gradient-based learning applied to document recognition," Processings of the IEEE, pp. 2278-2324, 1998.
[12] D. Li, L.-Q. Liang and W.-J. Zhang, "Defect inspection and extraction of the mobile phone cover glass based on the principal components analysis," The International Journal of Advanced Manufacturing Technology, vol. 73, pp. 1605-1614, 2014.
[13] D. C. Cireşan, U. Meier, J. Masci, L. M. Gambardella and J. A. Schmidhuber, "Flexible, high performance convolutional neural networks for image classification," in Proceedings of the Twenty-Second International Joint Conference on Artificial Intelligence, Barcelona, Spain, 2011.
[14] J.-K. Park, B.-K. Kwon, J.-H. Park and D.-J. Kang, "Machine learning-based imaging system for surface defect inspection," International Journal of Precision Engineering and Manufacturing-Green Technology, vol. 3, pp. 303-310, 2016.
[15] S. Faghih-Roohi, S. Hajizadeh, A. N´uñez, R. Babuska and B. D. Schutter, "Deep convolutional neural networks for detection of rail surface defects," in International Joint Conference on Neural Networks, Vancouver, Canada, 2016.
[16] W.-H. Lai and S.-Y. Tseng, "Defect classification of printed circuit board on deep convolutional neural networks," in Taiwan AOI Forum & Show, Hsinchu, Taiwan, 2017.
[17] C. Szegedy, A. Toshev and D. Erhan, "Deep Neural Networks for Object Detection," in Neural Information Processing Systems Conference, Lake Tahoe, Nevada, 2013.
[18] D. Scherer, A. Müller and S. Behnke, "Evaluation of Pooling Operations in Convolutional Architectures for Object Recognition," in International Conference on Artificial Neural Networks, Thessaloniki, Greece, 2011.
[19] V. Nair and G. E. Hinton, "Rectified linear units improve restricted boltzmann machines," in International Conference on International Conference on Machine Learning, Haifa, Israel, 2010.
[20] A. Krizhevsky, I. Sutskever and G. E. Hinton, "ImageNet classification with deep convolutional neural networks," in International Conference on Neural Information Processing Systems, Lake Tahoe, Nevada, 2012.
[21] "Very deep convolutional networks for large-scale image recognition," in International Conference on Learning Representations, San Diego, California, 2015.
[22] C. Szegedy, W. Liu, Y. Jia, P. Sermanet and S. Reed, "Going deeper with convolutions," in IEEE Conference on Computer Vision and Pattern Recognition, Boston, Massachusetts, 2015.
[23] M. Lin, Q. Chen and S. Yan, "Network in network," in International Conference on Learning Representations, Banff, Canada, 2014.
[24] D. P. Kingma and J. L. Ba, "Adam: a method for stochastic optimization," in International Conference on Learning Representations, San Diego, California, 2015.
[25] Y. Nesterov, "A method for solving the convex programming problem with convergence rate O (1/k^ 2)," Dokl. akad. nauk Sssr, vol. 269, pp. 543-547, 1983.
[26] G. E. Hinton, N. Srivastava and K. Swersky, "Neural Networks for Machine Learning - Lecture 6a - Overview of mini-batch gradient descent," 2012. [Online]. Available: https://class.coursera.org/neuralnets-2012-001/lecture.