進階搜尋


   電子論文尚未授權公開,紙本請查館藏目錄
(※如查詢不到或館藏狀況顯示「閉架不公開」,表示該本論文不在書庫,無法取用。)
系統識別號 U0026-0208201911043400
論文名稱(中文) 應用影像特徵分析於擠製成型視覺檢測之研究
論文名稱(英文) The Study of Vision Inspection on Material Extrusion Process with Image Feature Analysis
校院名稱 成功大學
系所名稱(中) 航空太空工程學系
系所名稱(英) Department of Aeronautics & Astronautics
學年度 107
學期 2
出版年 108
研究生(中文) 李柏翰
研究生(英文) Bo-Han Lee
學號 P46061241
學位類別 碩士
語文別 中文
論文頁數 105頁
口試委員 指導教授-賴維祥
口試委員-楊憲東
口試委員-許毅然
中文關鍵字 積層製造  擠製成型  自動化光學檢測  影像特徵  機器學習 
英文關鍵字 Additive Manufacturing  Material Extrusion  Automated Optical Inspection  Image Features  Machine Learning 
學科別分類
中文摘要 材料擠製成型(Material Extrusion, ME)屬於積層製造(Additive Manufacturing, AM)技術的一種,也是使用上最為普及;這個製程是藉由小孔徑的熱熔性材料的擠出,以規劃路徑層狀堆疊成型。然而過程中一有瑕疵產生就會隨著後續的列印逐漸放大致使整體列印失敗。因此本論文主要目標為加入自動化光學檢測的方法,使實驗室內的擠製成型三維列印機能自動識別當下列印狀況並因應作動,達到製程自動化的效果。
論文中利用PYCHARM 2018環境建立PYQT人機介面視窗來整合整個檢測系統。通訊方面,以微處理器為主做兩端的雙向通訊,一端以USB序列埠連結三維列印機完成G-code的字串傳輸,另一端以TCP/IP傳輸影像至人機介面端。演算法方面,本研究以樣板追蹤噴頭的方式定義感興趣區域(Region of Interest, ROI)以作為檢測的單位,並計算直方圖以及灰階共生矩陣的影像特徵以建立數據資料庫。資料庫內的資料用來訓練與評估支持向量機(Support Vector Machine, SVM)模型,最終部屬至人機介面以建立影像辨識系統。實驗結果顯示系統能夠完全檢測出深色消光線材的列印狀況(成功列印以及堆疊失敗狀態),並且針對無擠出狀態也能在噴頭與列印元件高度差約7至10mm時完全檢測出來。
英文摘要 Material Extrusion (ME) is one of processes for additive manufacturing, and it’s the most popular one. The definition of this process is heating plastic material and then extruding from 0.4mm nozzle to constantly stack layer by layer so that the shape designed in advance can be formed. However, if there is a flaw during the process, it’ll get bigger than bigger and then end up with printing failure.
Therefore, the main goal for this thesis is to solve this problem for ME machinery with automated optical inspection (AOI) system. The process is that camera captures the front field of view first. After that, the human machine interface (HMI) will show the image processed by the algorithm designed to track the nozzle’s position and define the region of interest as known as ROI, which would be extracted the image features to build the databases. When the databases are large enough, It shall use it to train SVM models as the classifier deployed on the HMI to identify if the component produced during the manufacturing time is defective or not and then transfer the corresponding G-code to make the 3D printer automated.
The results of the experiment in this research show that the AOI system can completely detect the production situation (success and stack failure) for 3D printer with dark material PLA, also is capable to detect totally the non-extrusion state with a condition that the difference between the height of the nozzle and the printed component is about 7 to 10 mm.
論文目次 中文摘要 i
英文摘要 ii
誌謝 v
目錄 vi
表目錄 x
圖目錄 xi
符號表 xv
第一章 緒論 1
1-1 前言 1
1-2 研究動機與目的 2
1-3 文獻探討 4
1-4 研究架構與方法 8
第二章 自動化光學檢測系統建立 10
2-1 機器視覺簡介 10
2-2 實驗設備與架構 12
2-2-1 硬體設備 13
2-2-2 軟體系統 19
2-3 電腦視覺演算法介紹 21
2-3-1 影像前處理 21
2-3-2 樣板演算法 24
2-3-3 影像特徵萃取 26
2-4 機器學習演算法介紹 40
2-4-1 機器學習 40
2-4-2 學習流程 40
2-4-3支持向量機演算法 41
第三章 影像辨識系統 46
3-1 系統概述 47
3-2 影像處理分析 48
3-2-1 噴頭追蹤與擷取ROI 49
3-2-2 特徵計算計算與分析 56
3-3 建立影像特徵資料庫 59
3-3-1 影像蒐集 59
3-3-2 影像特徵資料庫之建立 61
3-4 決策機制 62
3-4-1 建立決策機制的原因 62
3-4-2 決策機制架構 64
第四章 機器學習模型訓練與評估 67
4-1 特徵資料分析 67
4-1-1 資料外洩評估 68
4-1-2 標籤編碼 70
4-2 模型訓練流程及評估 70
4-2-1機器學習模型訓練流程 70
4-2-2 機器學習模型評估 72
4-3訓練模型分析 79
4-3-1 HOG-SVM架構之模型評估 80
4-3-2 SVM模型評估 83
4-3-3 ROC曲線評估 86
4-3-4綜合評估 87
4-4 特徵資料差異性評估與分析 88
4-4-1 各特徵資料庫模型訓練評估 89
4-4-2 探討直方圖變異數特徵是否影響模型訓練 89
4-4-3 探討直方圖眾數特徵是否影響模型訓練 91
第五章 實驗結果與討論 93
5-1 實驗模型 93
5-2 測試實驗設計 94
5-2-1正常列印的試驗方法 94
5-2-2堆疊失敗的試驗方法 95
5-2-3無擠出狀態的試驗方法 97
5-3 實驗結果分析與評估 98
第六章 結論與未來展望 100
6-1 結論 100
6-2 未來展望 101
參考文獻 102

參考文獻 [1] R. Minetoo, N. Volpato, J. Stolfi, R. M.M.H. Gregori, M. V.G.da Silva, “An Optimal Algorithm for 3D Triangle Mesh Slicing,” Federal University of Technology, 2015.
[2] A. Kumar, IEEE, “Computer-Vision-Based Fabric Defect Detection: A Survey,” IEEE Transactions on Industrial Electronics, Vol.55, No.1, 2008.
[3] H. Y.T Ngan, G. K.H Pang, N. H.C Yung, “Automated Fabric Defect Detection – A Review,” Image and Vision Computing 29, pp.442-458, 2011.
[4] J. Jiang, X. Xu, J. Stringer, “Support Structures for Additive Manufacturing : A Review,” University of Auckland, 2018.
[5] P. Chennakesava, Y. S. Narayan, “Fused Deposition Modeling – Insight,” International Conference on Advances in Design and Manufacturing, 2014.
[6] I. Zein, W. Hutmacher, K. C. Tan, S. Teoh, “Fused Deposition Modeling of Novel Scaffold Architectures for Tissue Engineering Applications,” Biomaterials, Vol.23, pp.1169-1185, 2002.
[7] A. Alafaghani, A. Qattawi, B. Alrawi, A. Guzman, “Experimental Optimization of Fused Deposition Modelling Processing Parameters : a Design-for-Manufacturing Approach,” 45th SME North American Manufacturing Research Conference, 2017.
[8] A. Garg, A. Bhattacharya, “An Insight to the Failure of FDM Parts Under Tensile Loading : Finite Element Analysis and Experiment Study,” International Journal of Mechanical Sciences 120, pp.225–236, 2017.
[9] F. Baumann, D. Roller, “Vision Based Error Detection for 3D Printing Process,” ICFST, 2016.
[10] 陳治戎, 使用HOG-SVM架構的交通標誌辨識, 國立交通大學, 2016.
[11] M.J. Tsai, S.D. Chiou, “Recognition of IC Chips via Machine Vision,” Journal of the Chinese Institute of Engineers, Vol.18, No.3, pp. 397-409, 1995.
[12] E.K. Teoh, D.P. Mital, B.W. Lee, L.K. Wee, “Automated Visual Inspection of Surface Mount PCBs,” University of Nanyang Technological, 1990.
[13] D.X. Li, Senior Member, H. Wu, IEEE, S. Li, “Internet of Things in Industries : A Survey,” IEEE, Vol.10, No.4, pp.2233-2243, 2014.
[14] L. Mosner, “Building Delta Style FDM 3D Printer,” Excel@FIT, 2015.
[15] W. K. Pratt, Digital Image Processing, Fourth Edition, A John Wiley & Sons Inc. Publication, 2007.
[16] W. K. Pratt, “Correlation Techniques of Image Registration,” IEEE Trans. on Aerospace and Electronic Systems, Vol.AES-10, pp.353-358, 1974.
[17] J. P. Lewis, “Fast Template Matching, ”Vision Interface95, Quebec City, Canada, 15-19 May, pp.120-123, 1995.
[18] S. Ourselin, X. Pennec, R. Stefanescu, X. Pennec, R. Stefanescu, “Robust Registration of Multi-Modal Medical Images : Towards Real-Time Clinical Applications,” 2001.
[19] M.B. Hisham, S. N. Yaakob, A. Nazren, N.M. W. Embedded, “Template Matching Using Sum of Squared Difference and Normalized Cross Correlation,” University Malaysia Perlis, 2015.
[20] Q. R. Razlighi and N. Kehtarnavaz, “A Comparison Study of Image Spatial Entropy,” University of Texas at Dallas, 2009.
[21] N. Dalal and B. Triggs, “Histograms of Oriented Gradients for Human Detection, ”IEEE Computer Society Conference, Computer Vision and Pattern Recognition, Vol.1, pp.886-893, 2015.
[22] P.Y. Chen, C.C. Huang, C.Y. Lien, and Y.H. Tsai, “An Efficient Hardware Implementation of HOG Feature Extraction for Human Detection,” in Proc.IEEE Computer Vision and Pattern Recognition(CVPR), Vol.1, pp.886-893, 2005.
[23] R. M. Haralick, K. Shanmugam, I. Dinstein, “Textural Features for Image Classification,” IEEE Transaction on Systems, Vol.SMC-3, No.6, 1973.
[24] H.T. Lin, S.A.M. Yaser, M.I. Malik, “Learning from Data,” AMLbook, 2012.
[25] I. V. Tetko, D. J. Livingstore, and A. I. Luid, “Neural Network Studies. 1. Comparison of Overfitting and Overtraining,” Compute. Sci., No.35, pp.826-833, 1995.
[26] R. Kohavi, “A Study of Cross-Validation and Bootstrap for Accuracy Estimation and Model Selection,” International Joint Conference on Artificial Intelligence, 1995.
[27] C. Cortes, V. Vapnik, “Support Vector Network,” Machine Learning, Vol.20, pp.273-297, 1995.
[28] B.E Boser, I. M. Guyon, and V. Vapnik, “A Training Algorithm for Optimal Margin Classifiers,” Fifth Annual Workshop on Computational Learining Theory ACM, pp.144-152, 1992.
[29]https://zh.wikipedia.org/wiki/%E6%94%AF%E6%8C%81%E5%90%91%E9%87%8F%E6%9C%BA
[30] J. H. Min and Y. Lee, “Bankruptcy Prediction Using Support Vector Machine with Optimal Choice of Kernel Function Parameters,” Expert Syst.Appl., Vol.28, pp.603-614, 2005.
[31] C. W. Hsu, C. J. Lin, “A Comparison of Methods for Multiclass Support Vector Machines,” IEEE Transactions on Neural Networks, Vol.13, pp.415-425, 2002.
[32] B. Wang, Z.W. Li, M.J. Li, “Efficient Duplicate Image Detection Algorithm for Web Images and Large-scale Database,” Technical Report of Microsoft Research, 2005.
[33] https://en.wikipedia.org/wiki/One-hot
[34] T. Fawcett, “An Introduction to ROC Analysis,” Pattern Recognition Letters, pp.861-874, 2006
[35]https://scikit-learn.org/stable/auto_examples/model_selection/plot_roc.html
[36] https://zh.wikipedia.org/wiki/ROC%E6%9B%B2%E7%BA%BF
論文全文使用權限
  • 同意授權校內瀏覽/列印電子全文服務,於2023-08-01起公開。
  • 同意授權校外瀏覽/列印電子全文服務,於2023-08-01起公開。


  • 如您有疑問,請聯絡圖書館
    聯絡電話:(06)2757575#65773
    聯絡E-mail:etds@email.ncku.edu.tw