在銑削製程中，刀具磨損缺陷會影響工件品質，當採取以影像識別刀具磨耗時，當擷取範圍過大，常易因金屬反射關係，致使缺陷區域影像不完整；當擷取範圍小，相同缺陷區域在不同影像上存在部分重疊的問題。此外，典型二維影像邊緣檢測方法亦會造成出磨耗輪廓的破壞，更缺少缺陷的深度資訊。目前三維檢測儀器成本昂貴，對於缺陷區域也缺乏可視化的直觀效果。
本研究首先提出刀具缺陷檢測之強化方法，並重建刀具缺陷的三維圖形。在金屬反射處理上，藉由多角度的檢測燈源設計，分別擷取各刃最大磨耗範圍，可組合出各刃磨耗區域的強化影像。在邊緣檢測上，透過Sobel與Canny方法進行磨耗邊界補償，可擷取出完整的刀具磨耗區域。在深度計算上，藉由點雲的三維圖形重建，比對新刀與磨耗刀的體積差，可獲知各刃的磨耗體積。最後，在可視化上，結合深度顏色對磨耗刀進行渲染，可直觀判斷出各刃磨耗區域深度。
在研究成果上，對於刀具不同刃上的磨耗缺陷，所提出的缺陷檢測法可獲得相當完整的磨耗區域檢視能力。對於刀具缺陷的三維圖形重建，以某一磨耗刀為例，最大體積誤差百分比在0.247%，本研究更能以顏色渲染出因磨耗所構成的深度差，從而呈現可視化的三維缺陷。

In milling processes, tool wear defects can affect the quality of the workpiece. When using image recognition to identify tool wear, capturing a large area often leads to incomplete defect images due to metal reflections. On the other hand, capturing a smaller area may result in partially overlapping defect regions in different images. Additionally, typical 2D edge detection methods can cause damage to the worn contour and lack depth information about defects. Currently, 3D inspection instruments are expensive, and there is a lack of intuitive visualization for defect areas.
This study proposes an enhanced method for tool wear defect detection and reconstructs the 3D representation of the tool wear defects. For handling metal reflections, a multi-angle detection light source design is employed to capture the maximum wear range of each cutting edge, enabling the creation of enhanced images for each blade's wear area. Edge detection using the Sobel and Canny methods compensates for wear boundary problems, allowing for complete extraction of the tool wear regions. The depth calculation is achieved through 3D reconstruction of point clouds, and by comparing the volume difference between a new tool and a worn tool, the wear volume of each blade can be determined. Finally, visualization is achieved by rendering the worn tool using depth color, enabling intuitive evaluation of the depth of wear areas for each blade.
The research results demonstrate that the proposed defect detection method offers a comprehensive view of wear areas on different tool edges. The 3D reconstruction of tool wear defects, using a specific worn tool as an example, exhibits a maximum volume error percentage of 0.247%. Moreover, this research enables color rendering of the depth differences caused by wear, providing a visual representation of 3D defects.

摘要 i
Abstract ii
誌謝 iii
目錄 iv
表目錄 vi
圖目錄 vii
附錄 ix
變數定義 x
中英文名詞對照表 xi
第1章 緒論 1
1.1 研究背景 1
1.2 研究目的 7
1.3 研究架構 9
第2章 理論方法 10
2.1 二維影像處理 10
2.1.1 噪聲濾除 10
2.1.2 邊緣檢測 11
2.1.3 洪水填充法(Flood-fill) 13
2.2 點雲處理 14
2.2.1 剛性變換(Rigid Transformation) 14
2.2.2 迭代最近點（Iterative Closest Point, ICP） 14
2.2.3 感興趣區域(Region Of Interest，ROI) 15
2.2.4 點雲表面建立 15
2.3 三維圖形處理 18
2.3.1 徑向基函數(Radial Basis Function, RBF) 18
2.3.2 四面體體積 20
2.3.3 高斯散度定理(Gauss's Divergence Theorem) 21
第3章 系統開發 22
3.1 系統架構 22
3.1.1 二維影像拍攝系統流程 24
3.1.2 三維圖形建立系統流程 24
3.2 刀具二維影像處理之改善方法 27
3.2.1 刀具影像前處理 28
3.2.2 刀具最大磨耗面積 30
3.3 刀具三維磨耗體積計算方法 31
3.3.1 點雲處理 31
3.3.2 體積計算方法 38
3.4 刀具磨耗深度計算方法 39
第4章 案例研究與實驗分析 40
4.1 刀具磨耗面積計算 40
4.1.1 機台刀具拍攝 40
4.1.2 刀具磨耗面積 41
4.2 刀具磨耗體積計算 46
4.2.1 刀具點雲數據提取 46
4.2.2 刀具點雲表面建立 47
4.2.3 刀具體積計算 59
4.2.4 刀具體積計算結果比較 61
4.3 刀具磨耗可視化 64
4.3.1 刀具深度渲染 64
第5章 結論與未來展望 67
5.1 結論 67
5.2 未來展望 69
參考文獻 70
附錄 72

[1]P. J. Bagga, M. A. Makhesana, and K. M. Patel, “A novel approach of combined edge detection and segmentation for tool wear measurement in machining,’’ Production Engineering, Vol. 15, pp. 519­533, 2021.
[2]O. G. Moldovan, S. Dzitac, I. Moga, T. Vesselenyi, and I. Dzitac, “Tool-wear analysis using image processing of the tool flank,’’ Symmetry, Vol. 9, Issue. 12, pp. 296, 2017.
[3]J. Yu and J. Zhou, “A Machine vision method for non-contact Tool Wear Inspection,’’ in 2020 Chinese automation congress (CAC) IEEE. pp. 1239­1243, November, 2020.
[4]G. Kiswanto, R. G. Putra, M. R. Fitriawan, F. Hiltansyah, and S. K. Putri, “Initial Development of Image Processing Based Computer Vision Technology on Robotic Arm Manipulator for Tool Wear Monitoring on Micro-milling,” in 2021 7th International Conference on Mechatronics and Robotics Engineering (ICMRE) IEEE, pp. 22­28, February, 2021.
[5]P. Jain and V. Tyagi, “A survey of edge-preserving image denoising methods,” Information Systems Frontiers, Vol. 18, pp. 159­170, 2016.
[6]Y. Wang, B. Zou, J. Wang, Y. Wu, and C. Huang, “Effect of the progressive tool wear on surface topography and chip formation in micro-milling of Ti–6Al–4V using Ti (C7N3)-based cermet micro-mill,” Tribology International, Vol. 141, 2020.
[7]X. Liang, Z. Liu, G. Yao, B. Wang, and X. Ren, “Investigation of surface topography and its deterioration resulting from tool wear evolution when dry turning of titanium alloy Ti-6Al-4V,” Tribology international, Vol. 135, pp. 130­142. 2019.
[8]M. A. Kuttolamadom, M. Laine Mears, T. R. Kurfess, U. Burger, and A. Bryan, “On the volumetric assessment of tool wear in machining inserts with complex geometries—part II: experimental investigation and validation on Ti-6Al-4V,” Journal of manufacturing science and engineering, Vol. 134, Issue. 5, 2012.
[9]T. G. Dawson and T. R. Kurfess, “Quantification of tool wear using white light interferometry and three-dimensional computational metrology,” International Journal of Machine Tools and Manufacture, Vol. 45, Issue. 4­5, pp. 591­596, 2005.
[10]https://zh.wikipedia.org/zh-tw/Flood_fill. July, 2023.
[11]https://www.cnblogs.com/shushen/p/5759679.html. July, 2023.
[12]https://mathworld.wolfram.com/Tetrahedron.html, July, 2023.
[13]https://nl.mathworks.com/help/matlab/ref/view.html, July, 2023.
[14]E. Kreyszig, “Vettor Integral Calculus. Integral Theorems,” Advanced Engineering Mathematics, 9th ed. John Wiley and Sons, 2007, Ch. 10, p. 458­460.
[15]P. J. Besl and N. D. McKay, “Method for registration of 3-D shapes,” in Sensor fusion IV: control paradigms and data structures, Vol. 1611, pp. 586­606, Spie. April, 1992.
[16]G. Turk and J. F. O'brien, “Shape transformation using variational implicit functions,” in ACM SIGGRAPH 2005 Courses, pp. 335­342, 2005.
[17]N. Amenta, M. Bern, and M. Kamvysselis, “A new Voronoi-based surface reconstruction algorithm,” in Proceedings of the 25th annual conference on Computer graphics and interactive techniques, pp. 415­421, July, 1998.
[18]M. Varnuska and I. Kolingerova, “Improvements to surface reconstruction by the CRUST algorithm,” in Proceedings of the 19th Spring Conference on Computer Graphics, pp. 89­97, April, 2003.
[19]W. Gao, X. Zhang, L. Yang, and H. Liu, “An improved Sobel edge detection,” in 2010 3rd International conference on computer science and information technology IEEE, Vol. 5, pp. 67­71, July, 2010.