近幾年德國提出工業4.0概念，各國開始投入製造智能化技術研發，如美國AMP先進製造、中國製造2025計畫、日本工業4.1J等，國內政府當前也跟隨推出智慧機械、生產力4.0等製造產業轉型計畫。在工具機部分，無人化產線已成當前智慧製造之趨勢，目前工具機製造產線雖已可達成自動化上下料與加工，但若是更換並未事前設定之工件，還是需要人力進行工件基準點校正，無法達成無人化，耗費時間成本較高；相對的，針對自動化基準點校正系統開發可縮減整體校正時間，且可使整體製造產線達成無人自動化之指標。在自動化基準點校正部分，德國DATRON[3]開發neo Measurement系統，藉由其自行開發之二維視覺系統可進行工件基準點自動校正，不過在操作上還是需要人員進行基準邊設定，在達到無人化目標上還是有差距。有鑑於此，本研究擬開發一套線上工具機基準點快速自動校正系統，將3D重建中的結構光掃描技術結合工具機控制應用於工具機產線生產製造上，並搭配自行開發改良之探針，降低整體設備成本，達成自動定位功能。研究主要分為四個部分，其一為以C++環境開發格雷編碼結構光掃描技術來重建出加工平台上工件三維空間資訊，找出工件實際空間座標；其二為透過PCL函式庫將重建出的三維點雲資料進行雜訊濾除、邊緣偵測等後處理取得所需之工件邊緣外型座標；其三是以自行開發之探測設備搭配重建出之工件邊緣模型進行自動工件原點校正；其四是以LabVIEW開發一套工具機線上即時監控與整合介面，將結構光掃描與自動化原點校正系統整合於該介面中。理論方面，本研究採用結構光掃描技術，圖形編碼採用格雷時序編碼，並搭配三角量測原理重建出加工平台工件點雲(Cloud Point)模型，並利用三維邊緣偵測進行邊緣掃描，將邊緣座標提取出來。第二部分則以自行開發訊號擷取模組搭配光電尋邊器進行工件邊緣找尋，利用工件點雲模型邊緣進行自動化工件原點定位。實驗結果顯示，以結構光掃描重建出的整體模型尺寸誤差在±0.1mm以內，透過偵測出的點雲邊緣座標進行定位的誤差浮動範圍在0.05mm以內；尋邊設備之間藍芽通訊所造成的時間延遲平均為11ms，造成的平均移動誤差為0.005mm；整體進行結構光掃描、點雲處理、點雲邊緣偵測與自動定位動作所耗費時間約為99~103秒，與傳統人為操作工件尋邊定位所耗費時間相比縮短很多並且達成無人自動化之目標。

Owing to introduce the concept of industry 4.0 from Germany. The various country began to invest in manufacturing intelligent technology research and development. United States proposed Advanced Manufacturing Partnership (ASP), China proposed Made in China 2025, Japanese proposed Industry 4.1J. The Taiwan government also currently introduction of Smart Manufacturing and Taiwan Productivity 4.0 Initiative. In the machine tool section, the unmanned production line has become the current trend. Although the machine tool manufacturing production line has been able to achieve automatic loading and unloading for the workpiece. However, if replace the workpiece that has not been set in advance. It still required to manually perform alignment for the workpiece. This will result in higher time and cost. Development for automated workpiece alignment system can reduce alignment time, and achieve the purpose of the unmanned factory. In the automated workpiece alignment section, Germany DATRON[3] develops the neo measurement system, which automatically aligns the workpiece by its self-developed 2D vision system. However, it still needs the manual operation.
Therefore, this study develops a 3D scanning system using the structured light of gray code for aiding the workpiece position of CNC machine tool online. The 3D scanning system automatically aids the machine tool probes for component setting and inspection. A 3D scanning system consisting of one camera and one projector with structural light has been set up in order to acquire the images representing the scanned object with projected light patterns. These images are then decoded and analyzed in order to obtain the depth information for each point representing the workpiece. The research is mainly divided into four parts. First, the workpiece is reconstructed 3D model points based on the structured light of gray code using C++ programming. Second, the noise filtering, edge detection, and other post-processing can obtain 3D model points using the Point Cloud Library (PCL). The 3D model points are large data sets composed of 3D physical point data. Third, to automate the workpiece alignment with the self-developed detection equipment and the reconstructed processing workpiece edge model. Fourth, we established a human-machine Interface to connect the CNC controller and the 3D scanning system for uploading data. Resulted showed that the dimensional accuracy error of the 3D scanning system after reconstructing 3D model points is about ±0.1mm for each axis. The repeated precision error of the 3D model edge points is about 0.05 mm. The time delay of Bluetooth communication between the edge-finding device is about 11 milliseconds, resulting average movement error is about 0.0005mm. The running time of the 3D scanning system with automatic workpiece position is about 99~103 seconds, including, structured light scanning, 3D model points processing, edge detection, and automatic positioning. Experiments show that this study can achieve the online fast automatic workpiece positioning on CNC machine tool.

摘要 ....i
英文摘要(Abstract) ....ii
誌謝 ....iv
目錄 ....v
表目錄 ....vii
圖目錄 ....viii
符號說明 ....xi
第 一 章 緒論 ....1
1.1 研究背景與動機 ....1
1.2 研究目的 ....2
1.3 論文架構 ....2
第 二 章 文獻回顧 ....3
2.1 3D重建原理分類 ....3
2.2 國內外相關研究 ....4
第 三 章 研究架構與方法 ....11
3.1 研究架構與流程 ....11
3.2 硬體及系統架構 ....12
3.2.1 結構光掃描硬體設備 ....12
3.2.2 工具機定位設備介紹 ....12
3.2.3 實驗機台介紹 ....14
3.3 相機成像與空間校正原理 ....15
3.3.1 相機成像原理 ....15
3.3.2 影像與世界座標系統轉換 ....16
3.3.3 影像校正 ....21
3.4 結構光掃描原理與建模處理 ....23
3.4.1 結構光原理 ....23
3.4.2 結構光編碼與解碼 ....24
3.4.3 結構光三角量測原理 ....29
3.4.4 點雲模型資料處理 ....31
3.5 工具機系統通訊與監控 ....36
3.5.1 PC-Based系統整合架構 ....36
3.5.2 新代控制器通訊連線 ....38
3.5.3 控制器PLC程式撰寫及控制 ....40
3.6 工具機定位輔助技術 ....41
3.6.1 工具機定位輔助設備 ....42
3.6.2 尋邊器與工具機通訊 ....43
3.6.3 工具機定位指令 ....44
第 四 章 實驗結果 ....46
4.1 結構光掃描建模系統 ....47
4.1.2 相機-投影機空間校正 ....49
4.1.3 虎鉗位置對於建模精度影響實驗 ....50
4.1.4 工件擺放位置對於建模精度影響實驗 ....53
4.1.5 重複建模尺寸精度實驗 ....56
4.2 工具機輔助自動定位系統 ....57
4.2.1 光電尋邊器單向定位精度 ....57
4.2.2 尋邊訊號延遲測試實驗 ....59
4.2.3 工具機自動定位精度實驗 ....61
第 五 章 結論與未來展望 ....64
附錄一 自動定位MACRO巨集指令 ....65
參考文獻 ....66
Extended Abstract ....68

[1] HEIDENHAIN 3D Touch Porbes,” https://www.heidenhain.com/en_US/products/touchprobes/”.
[2] RENISHAW Probes,” http://www.renishaw.com/en/machine-tool-probes-and-software--6073”.
[3] DATRON neo Measurement Probe,” https://www.datron.com/cnc-machineproducts/
probing.php”.
[4] CENTROID Touch Probes,” http://www.centroidcnc.com/cnc_touch_probe.html”.
[5] T. C. Strand,”Optical three-dimensional sensing for machine vision”, Optical Engineering 24(1), 033-040, 1985.
[6] Hao Li,”CSCI 621:Digital Geometry Processing”, Institute for Creative Technology, University of Southern California.
[7] S. Garrido-Jurado, R. Muñoz-Salinas, F.J. Madrid-Cuevas, M.J. Marín- Jiménez, ”Simultaneous reconstruction and calibration for multi-view structured light scanning”. Journal of Visual Commuincation and Image Representation, Pages 120-131, August 2016.
[8] Changsoo Je, Kwang Hee Lee, Sang Wook Lee, “Multi-projector color struceured-light vision”, Signal Processing: Image Communication, Pages 1046-1058, October 2013.
[9] Jin Lu, Rong Mo, Hubin Sun, Zhiyong Change, Xiaxia Zhao, “Simplified absolute phase retrieval of dual-frequency fringe patterns in fringe projection profilometry”, Optics Communications, Pages 101-109, April 2016.
[10] Douglas Lanman, “Bulid Your Own 3D Scanner: 3D Photography for Begineers”, SIGGRAPH 2009 Course Notes.
[11] Christian Piccolo, “A LOW-COST, EFFICIENT AND ACCURATE HUMAN BODY SCANNER”, Dipartimento di Ingegneria Informatica , Università Degli Studi di Padova,.
[12] 陳昱佟，2016，”運用多週期結構光進行曲面輪廓量測及投影機校正”，國立中興大學機械工程研究所，碩士學位論文
[13] Hoang Dinh Cuong，”運用自動化機器手臂掃描與影像定位之三維表面數位量測研究”，國立台北科技大學自動化科技研究所，碩士學位論文
[14] Zhang Xi, Tian Xiaodong Yamazaki Kazuo, “On-machine 3D vision system for machining setup modeling”, Journal of Advanced Manufacturing Technology, Pages 251-265, 2010.
[15] Krzysztof Okarma, Marek Grudziński, “The 3D scanning system for the machine vision based positioning of workpieces on the CNC machine tools”, IEEE 2012 17th International Conference on Methods & Models in Automation & Robotics (MMAR).
[16] 李騏丞，2015，”立體視覺輔助路邊停車系統”，國立中山大學機械與機電工程學系，碩士學位論文
[17] 佘東穎，2014，”應用立體視覺3D 量測技術於五軸工具機防碰撞”，國立中大學電機研究所，碩士學位論文.
[18] 廖彥法，2005，”立體影像於物件幾何重建之研究”，國立成功大學航空太空工程研究所，碩士學位論文
[19] R. Tsai, “A versatile camera calibration technique for high-accuracy 3D machine vision metrology using off-the-shelf TV cameras and lenses”, IEEE Journal of Robotics and Automation, Pages 323-344, August 1987.
[20] Z. Zhang, “A Flexible New Technique for Camera Calibration”, Technical Report of Microsoft Research, MSR-TR-98-71.
[21] Jason Geng, “Structured-light 3D surface imaging: a tutorial”, IEEE Intelligent Transportation System Society, March 2011.
[22] Guido Gerig, “CS 6320-S2013: Structured light II”, Scientific Computing and Imageing Institute, University of UTAH, Spring 2013.
[23] Radu Bogdan Rusu, “Semantic 3D Object Maps for Everyday Manipulation in Human Living Environments”, Institut für Informatik , der Technischen Universität München
[24] Fengjun Hu, Yanwei Zhao, “DISCRETE POINT CLOUD FILTERING AND SEARCHING BASED ON VGSO ALGORITHM”, 27th Conference on Modelling and Sumulation, May 2013.
[25] Bastian Steder, Radu Bogdan Rusu, Kurt Konolige, Wolfram Burgard, “Point Feature Extraction on 3D Range Scans Taking into Account Object Boundaries”, IEEE International Conference on Robotics and Automation, May 2011.
[26] OpenCV, “https://opencv.org/”.
[27] Point Cloud Library(PCL), “https://pointclouds.org”.