臺灣博碩士論文加值系統

自動光學檢測在許多行業中得到了廣泛的應用，之前的品質管理主要是以人工目視檢測的方式，因應現今企業對產品的品質及生產效率愈加重視，此種人工的檢測方式也漸漸走向自動化。自動光學檢測設備是輔助企業導入自動化的檢測設備，主要應用面是對產品品質管理，此項技術在確保產品品質層面精度，可避免製造商在生產上的損失，非常符合客戶的需求。
本研究乃針對機車後視鏡支架表面瑕疵來做檢測，因當前檢測方式仍是以現場品管人員的目視檢測，但電鍍後之物體會呈現高反光的表面，檢測上需要耗費大量時間及人力，因此本研究提出結合影像辨識之深度神經網路的技術，希望藉以此研究開發出一套自動光學檢測系統，以取代人工目視檢查，實驗結果顯示，本研究所提出的方法能夠有效地提升檢測速度和具有較佳辨識率。

Automatic optical inspection has been widely used in many industries. In the past, quality management was mainly in the form of manual visual inspection. In response to the more concern of today's enterprises on product quality and production efficiency, this manual testing method is gradually moving towards automation. Automatic optical inspection equipment is an auxiliary enterprise to import automated detection equipment, the main application surface is product quality management testing. The main application area is product quality management. This technology ensures the accuracy of product quality, avoids the loss of manufacturers in production, and allows products to meet buyers' quality requirements.
This study proposed a method to detect defects on the surface of the motorcycle rearview mirror bracket. Because the current inspection method is still the visual inspection of the on-site quality control personnel, the electroplated object will show a highly reflective surface, so the inspection requires a lot of time and manpower. This study proposed a combination of image recognition depth neural network technology to develop a set of research automatic optical inspection system to replace manual visual inspection. The experimental results show that the method proposed in this study can effectively improve the detection speed and have a better identification rate.

摘要 i
Abstract ii
誌謝 iv
目錄 v
圖目錄 viii
表目錄 xi
第一章、 緒論 1
1.1 研究背景及動機 1
1.2 文獻探討 3
1.3 論文架構 9
第二章、 相關背景知識之探討 10
2.1 自動光學檢測 10
2.2 人工智慧(Artificial Intelligence) 12
2.2.1 機器學習(Machine Learning) 12
2.2.2 深度學習(Deep Learning) 13
2.3 資料擴增(Data Augmentation) 15
2.4 後照鏡支架製程及檢驗標準 17
2.4.1 後照鏡支架製程介紹 17
2.4.2 鏡骨鍍鉻外觀檢驗標準 23
第三章、 系統設計 28
3.1 自動光學檢測設備及架設 30
3.2 系統檢測流程 34
3.3 系統模型及資料庫建立 36
第四章、 實驗分析 42
4.1 實驗環境 42
4.2 實驗結果 43
4.3 實驗比較 47
4.3.1 系統驗證方法 47
4.3.2 驗證結果及比較 49
第五章、 結論與建議 51
5.1 結論 51
5.2 後續延伸研究 51
參考文獻 52

[1]丁于倢（2019）。基於卷積神經網絡之微小探針表面剝金瑕疵自動光學檢測。國立臺北大學資訊工程學系碩士論文，新北市。 取自https://hdl.handle.net/11296/3cd5bc
[2]羅倆凡（2019）。利用具注意力機制之生成對抗網路於工業影像之瑕疵單一分類。國立交通大學電控工程研究所碩士論文，新竹市。 取自https://hdl.handle.net/11296/4r837a
[3]林靜甫（2018）。鏡面不銹鋼板瑕疵自動光學檢測系統之研製。國立高雄應用科技大學機械工程系碩士論文，高雄市。 取自https://hdl.handle.net/11296/ge3759
[4]蘇詠靖 ； 游原瑋 ； 何昭慶（2018）。結合機器視覺與深度學習之金屬圓柱表面缺陷檢測系統。科儀新知， P46 – 58
[5]楊福生（2020）。基於機器視覺及光度立體技術應用於橡膠墊圈之自動光學檢測系統。國立臺北科技大學製造科技研究所碩士論文，台北市。 取自https://hdl.handle.net/11296/jb4y5y
[6]倪瑞宏（2019）。自動光學檢測技術應用於磁粉探傷之研究。國立高雄科技大學機械工程系碩士論文，高雄市。 取自https://hdl.handle.net/11296/mr5a3j
[7]張峻睿（2019）。隱形眼鏡乾片圖紋瑕疵檢測系統。國立雲林科技大學電機工程系碩士論文，雲林縣。 取自https://hdl.handle.net/11296/r4479t
[8]S. Niu, B. Li, X. Wang and H. Lin, "Defect Image Sample Generation With GAN for Improving Defect Recognition," in IEEE Transactions on Automation Science and Engineering, vol. 17, no. 3, pp. 1611-1622, July 2020, doi: 10.1109/TASE.2020.2967415.
[9]J. O’Leary, K. Sawlani and A. Mesbah, "Deep Learning for Classification of the Chemical Composition of Particle Defects on Semiconductor Wafers," in IEEE Transactions on Semiconductor Manufacturing, vol. 33, no. 1, pp. 72-85, Feb. 2020, doi: 10.1109/TSM.2019.2963656.
[10]Q. Xie, D. Li, J. Xu, Z. Yu and J. Wang, "Automatic Detection and Classification of Sewer Defects via Hierarchical Deep Learning," in IEEE Transactions on Automation Science and Engineering, vol. 16, no. 4, pp. 1836-1847, Oct. 2019, doi: 10.1109/TASE.2019.2900170.
[11]W. Wang, S. Chen, L. Chen and W. Chang, "A Machine Vision Based Automatic Optical Inspection System for Measuring Drilling Quality of Printed Circuit Boards," in IEEE Access, vol. 5, pp. 10817-10833, 2017, doi: 10.1109/ACCESS.2016.2631658.
[12]C. L. S. C. Fonseka and J. A. K. S. Jayasinghe, "Implementation of an Automatic Optical Inspection System for Solder Quality Classification of THT Solder Joints," in IEEE Transactions on Components, Packaging and Manufacturing Technology, vol. 9, no. 2, pp. 353-366, Feb. 2019, doi: 10.1109/TCPMT.2018.2864666.
[13]Simonyan, Karen, and Andrew Zisserman. "Very deep convolutional networks for large-scale image recognition." arXiv preprint arXiv:1409.1556 (2014).
[14]He, Kaiming, et al. "Deep residual learning for image recognition." Proceedings of the IEEE conference on computer vision and pattern recognition. 2016.
[15]Szegedy, Christian, et al. "Rethinking the inception architecture for computer vision." Proceedings of the IEEE conference on computer vision and pattern recognition. 2016.