Responsive image
博碩士論文 etd-0716121-152433 詳細資訊
Title page for etd-0716121-152433
論文名稱
Title
基於形態學卷積神經網路的銀膠印刷電路缺陷偵測
Silver Glue Printed Circuit Defect Detection based on Morphological Convolutional Neural Networks
系所名稱
Department
畢業學年期
Year, semester
語文別
Language
學位類別
Degree
頁數
Number of pages
40
研究生
Author
指導教授
Advisor
召集委員
Convenor
口試委員
Advisory Committee
口試日期
Date of Exam
2021-07-02
繳交日期
Date of Submission
2021-08-16
關鍵字
Keywords
卷積神經網路、人工智慧、視覺、機器視覺、工業自動化
Convolution Neural Network, CNN, Industrial automation, vision, Artificial Intelligence
統計
Statistics
本論文已被瀏覽 372 次,被下載 0
The thesis/dissertation has been browsed 372 times, has been downloaded 0 times.
中文摘要
對於工業自動化來說,雖然目前導入了許多人工智慧(Artificial Intelligence ,AI),其實並非每項應用都有使用AI,尤其在工業自動化上機器視覺的部分,沒有那麼普及化,多數採形態學分析影像識別結果,運用形態學影像系統在各層面表現的確穩定,但是功能越強大,參數就越多,即不便於使用端調整,參數固定後對於環境變化接受度低,對現在科技產業的產品型態少量多樣實況而言,調整上確沒那麼方便,以AI應用來說,除了程式本身思考邏輯不同,以及特徵提取的方式不同以外,在特定應用上使用方便許多,相對少了許多參數,也較能克服環境的變化,可為使用者帶來工業上便捷的應用,此研究以銀膠為例,因為膠體的特性形狀並非那麼固定,藉由AI模型辨識來識別短路或缺陷型態,進而達到工業應用的需求。

本篇論文提出卷積神經網路(Convolutional Neural Network ,CNN)方式來識別銀膠狀態,以區別短路或缺陷,以CNN的方式來做出研究,透過不一樣的層數模型架構實驗,並且觀察實驗數據,以達到學習以及研究的效果,與現有機器上視覺系統做比較,並歸納出兩者差異性,透過此實驗可以延伸觀念於應用於產業上,來達到產業技術創新的目的,並且朝向應用面做發展。
Abstract
For Industrial automation environment , although use more Artificial Intelligence(AI) application ,but not all , especially it use on industrial vision system is seldom, Most vision system is use morphology to analysis , it is stable in morphology vision system ,but the system is more powerful , the more difficult to set parameter because has more parameter amount, it is inconvenient to operator ,but after system parameter set , it is lower adaptable to the environment , for the currently product type of technology is small volume and large variety production , it is inconvenient to adjustment , in terms of AI ,besides program logic and feature get method ,parameter amount is different ,AI has less parameter set and higher adaptable to the environment ,it is convenient in the industrial environment to application ,this research is silver glue printed circuit for example , use AI to analysis circuit loss type and short type , to achieve industrial application.
This research is use Convolution Neural Network(CNN) to analysis circuit loss type and short type ,it is use Lenet neural network structure to do , although Lenet is a earlier model ,but this research use different amount layer to structure model , and observe experimental data to achieve learn and research , through this research extended idea to industrial application , and creation a new technology in the industry , achieve to application develop.
目次 Table of Contents
目 錄
論文審定書 i
誌 謝 ii
摘 要 iii
Abstract iv
目 錄 v
圖 次 vii
表 次 viii
第一章 緒論 1
1.1 研究背景 1
1.2 研究動機 1
1.3 研究目的 2
第二章 文獻探討 3
2.1 形態學Morphology 3
2.2機器學習 Machine Learning 4
2.3 深度學習 Deep Learning 4
2.4 卷積神經網路 Convolutional Neural Network 5 2.5 Multi Classification learning 6 2.6 Multi Label learning 7
2.7 Gradient-weighted Class Activation Mapping(Grad CAM) 8
2.8 Local Interpretable Model-agnostic Explanations(LIME) 9
第三章 研究方法與步驟 10
3.1 研究方法 10
3.1.1 資料獲取 11
3.1.2 架構卷積神經網路(CNN) 11
3.1.3 模型輸出分類 12
3.2 模型解釋與評估 12
3.2.1 模型解釋 12
3.2.2 模型評估 13
3.3 研究架構 13
第四章 實驗結果與討論分析 14
4.1資料整理 14
4.2 研究流程 14
4.3 研究過程 15
4.3.1 模型訓練 15
4.3.2 預測結果 16
4.3.3 模型解釋 18
4.3.4 模型評估 20
4.4 研究分析 21
第五章 研究結論與建議 22
5.1 研究結論 22
第六章 參考文獻 23
附錄A 25
參考文獻 References
[1] A. G. Hanbury and J. Serra, “Morphological operators on the unit circle,” IEEE Trans. Image Process., vol. 10, no. 12, pp. 1842–1850, Dec. 2001, doi: 10.1109/83.974569.
[2] P. Maragos and R. Schafer, “Morphological filters--Part I: Their set-theoretic analysis and relations to linear shift-invariant filters,” IEEE Trans. Acoust. Speech Signal Process., vol. 35, no. 8, pp. 1153–1169, Aug. 1987, doi: 10.1109/TASSP.1987.1165259.
[3] Y. LeCun, Y. Bengio, and G. Hinton, “Deep learning,” Nature, vol. 521, no. 7553, pp. 436–444, May 2015, doi: 10.1038/nature14539.
[4] Y. LeCun, L. Bottou, Y. Bengio, and P. Ha, “Gradient-Based Learning Applied to Document Recognition,” p. 46, 1998.
[5] F. Sultana, A. Sufian, and P. Dutta, “Advancements in Image Classification using Convolutional Neural Network,” 2018 Fourth Int. Conf. Res. Comput. Intell. Commun. Netw. ICRCICN, pp. 122–129, Nov. 2018, doi: 10.1109/ICRCICN.2018.8718718.
[6] M. Sahare and H. Gupta, “A Review of Multi-Class Classification for Imbalanced Data,” Int. J. Adv. Comput. Res., vol. 2, no. 3, p. 6.
[7] M. Aly, “Survey on Multiclass Classification Methods,” p. 9.
[8] A. Vallet and H. Sakamoto, “A Multi-Label Convolutional Neural Network for Automatic Image Annotation,” J. Inf. Process., vol. 23, no. 6, pp. 767–775, 2015, doi: 10.2197/ipsjjip.23.767.
[9] M. S. Sorower, “A Literature Survey on Algorithms for Multi-label Learning,” p. 26.
[10] M. T. Ribeiro, S. Singh, and C. Guestrin, “‘Why Should I Trust You?’: Explaining the Predictions of Any Classifier,” ArXiv160204938 Cs Stat, Aug. 2016, Accessed: Jun. 14, 2021. [Online]. Available: http://arxiv.org/abs/1602.04938
[11] E. Spyromitros-Xioufis, W. Groves, G. Tsoumakas, and I. Vlahavas, “Multi-Label Classification Methods for Multi-Target Regression,” p. 10.
[12] M. Lin, Q. Chen, and S. Yan, “Network In Network,” ArXiv13124400 Cs, Mar. 2014, Accessed: Jun. 14, 2021. [Online]. Available: http://arxiv.org/abs/1312.4400
[13] P. Maragos and R. Schafer, “Morphological filters--Part I: Their set-theoretic analysis and relations to linear shift-invariant filters,” IEEE Trans. Acoust. Speech Signal Process., vol. 35, no. 8, pp. 1153–1169, Aug. 1987, doi: 10.1109/TASSP.1987.1165259.
[14] M. S. Sorower, “A Literature Survey on Algorithms for Multi-label Learning,” p. 26.
[15] H. N. Ghafil and D. M. B. Ali, “Cracks Measurement On The Basis Of Machine Vision,” vol. 16, no. 06, p. 7, 2016.
[16] Y. Kang, I.-L. Cheng, W. Mao, B. Kuo, and P.-J. Lee, “Towards Interpretable Deep Extreme Multi-Label Learning,” in 2019 IEEE 20th International Conference on Information Reuse and Integration for Data Science (IRI), Los Angeles, CA, USA, Jul. 2019, pp. 69–74. doi: 10.1109/IRI.2019.00024.
[17] A. M. Raid, W. M. Khedr, M. A. El-dosuky, and M. Aoud, “Image Restoration Based on Morphological Operations,” Int. J. Comput. Sci. Eng. Inf. Technol., vol. 4, no. 3, pp. 9–21, Jun. 2014, doi: 10.5121/ijcseit.2014.4302.
[18] A. Simon, M. S. Deo, S. Venkatesan, and D. R. R. Babu, “An Overview of Machine Learning and its Applications,” p. 4.
[19] A. Krizhevsky, I. Sutskever, and G. E. Hinton, “ImageNet classification with deep convolutional neural networks,” Commun. ACM, vol. 60, no. 6, pp. 84–90, May 2017, doi: 10.1145/3065386.
[20] R. R. Selvaraju, M. Cogswell, A. Das, R. Vedantam, D. Parikh, and D. Batra, “Grad-CAM: Visual Explanations From Deep Networks via Gradient-Based Localization,” p. 9.
[21] G. Madzarov, D. Gjorgjevikj, and I. Chorbev, “A Multi-class SVM Classifier Utilizing Binary Decision Tree,” p. 11.
[22] J. An, Y. Chen, and H. Shin, “Weather Classification using Convolutional Neural Networks,” in 2018 International SoC Design Conference (ISOCC), Daegu, Korea (South), Nov. 2018, pp. 245–246. doi: 10.1109/ISOCC.2018.8649921.

電子全文 Fulltext
本電子全文僅授權使用者為學術研究之目的,進行個人非營利性質之檢索、閱讀、列印。請遵守中華民國著作權法之相關規定,切勿任意重製、散佈、改作、轉貼、播送,以免觸法。
論文使用權限 Thesis access permission:自定論文開放時間 user define
開放時間 Available:
校內 Campus: 已公開 available
校外 Off-campus:開放下載的時間 available 2024-08-16

您的 IP(校外) 位址是 52.91.67.23
現在時間是 2024-03-28
論文校外開放下載的時間是 2024-08-16

Your IP address is 52.91.67.23
The current date is 2024-03-28
This thesis will be available to you on 2024-08-16.

紙本論文 Printed copies
紙本論文的公開資訊在102學年度以後相對較為完整。如果需要查詢101學年度以前的紙本論文公開資訊,請聯繫圖資處紙本論文服務櫃台。如有不便之處敬請見諒。
開放時間 available 2024-08-16

QR Code