近年來，因開發中國家人力成本高漲且歐美各國欲將製造業移回本土的主客觀條件，自動化的議題再次受到重視。同時，由於製造業產品逐漸走向客製化「少量多樣」的趨勢，使得產線的彈性應變能力比起以往更為關鍵。在現今的工業機器人應用中，表面加工佔了約三分之一，但若未能事先取得待加工物的三維模型，便很難規劃運動軌跡而影響加工精準度。一般多是透過人工教導的方式解決，但此法不僅耗時，準確度也相當受限。現今雖有基於雷射的逆向工程法可供選擇，但其設備不僅昂貴且系統設計之複雜度也較高，因此本論文規劃整合立體視覺於機械手臂解決上述問題，提升應用之便利性。但不可諱言的，多數之待加工物其明顯之特徵點過少，此特性將使傳統匹配演算法產生誤匹配進而影響三維重建的精度。有鑒於此，本論文發展了一個基於單應性矩陣之立體匹配法來改善此問題。本論文首先藉由結合多種特徵點偵測法以增加待加工物偵測上的特徵點數量，並透過幾個誤匹配排除方法以確保特徵點匹配的正確性。之後由此些特徵點估測一強健的單應性矩陣，並以此單應性矩陣計算匹配結果然後進行深度估測與三維重建。最後，將此精確之三維模型應用於具有高使用彈性的六軸關節型機械手臂並進行一個表面加工作業，以此整合製造業常見的CAD/CAM流程於同一平台進而提高應用之便利性。實驗結果顯示以此方法重建的三維模型之準確性佳，充分驗證此方法應用於表面加工的可行性。

In recent years, with labor costs in developing countries rapidly increasing and a trend in the manufacturing industry of moving back to America and Europe, the topic of automation has once again become a hot issue. In the meantime, the growing trend of customized “small-volume large-variety production” indicates that the flexibility of production line is more critical than ever. Nowadays, surface machining applications account for about one-third of applications in industrial robots. In this field, if a 3D geometry model of objects for machining cannot be obtained in advance, trajectory planning will be achieved only with much difficulty and the machining precision will be diminished. Conventionally, teach-by-showing is a solution, but it is time-consuming and the results are limited. Although laser-based inverse engineering approaches are available, generally these approaches are complicated and their costs are high. Therefore, this thesis plans to integrate the stereo vision into industrial robots to solve those problems. However, it is doubtless that most objects for machining have few features, so traditional stereo matching algorithms will fall short in reconstructing the 3D models of these objects. In order to deal with the aforementioned problem, this thesis develops a homography based stereo matching algorithm. Firstly, this thesis increases the number of features on objects for machining by combining several algorithms and then filters out the miss matches. Based on the correct matches, a robust homography matrix is estimated. Moreover, the stereo matching result is calculated through this matrix. Consequently, depth estimation and 3D reconstruction will be conducted. Several experiments have been conducted to assess the performance of the proposed approach. In the experiment, the reconstructed 3D model is used in a surface machining task performed by a six-axis articulated robot. Experimental results indicate that the approach proposed in this thesis provides good reconstructed results so as to verify the effectiveness of the proposed approach in surface machining.

[1] http://www.abb.com/cawp/gad00540/6fe355df884d914fc1257a23003903e2.aspx?single=1.
[2] J. A. Gangloff and M. F. de Mathelin, "Visual servoing of a 6-DOF manipulator for unknown 3-D profile following," IEEE Transactions on Robotics and Automation,vol. 18, pp. 511-520, 2002.
[3] W.-C. Chang, "Binocular vision-based 3-D trajectory following for autonomous robotic manipulation," Robotica, vol. 25, pp. 615-626, 2007.
[4] M. Dinham and G. Fang, "Autonomous weld seam identification and localisation using eye-in-hand stereo vision for robotic arc welding," Robotics and Computer-Integrated Manufacturing, vol. 29, pp. 288-301, 2013.
[5] T. Kanade and M. Okutomi, "A stereo matching algorithm with an adaptive window: Theory and experiment," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 16, pp. 920-932, 1994.
[6] J. Sun, N.-N. Zheng, and H.-Y. Shum, "Stereo matching using belief propagation," IEEE Transactions on Pattern Analysis and Machine Intelligence,vol. 25, pp. 787-800, 2003.
[7] A. Klaus, M. Sormann, and K. Karner, "Segment-based stereo matching using belief propagation and a self-adapting dissimilarity measure," in Proceedings of 18th International Conference on Pattern Recognition, ICPR 2006., pp. 15-18, 2006.
[8] X. Mei, X. Sun, M. Zhou, S. Jiao, H. Wang, and X. Zhang, "On building an accurate stereo matching system on graphics hardware," in Proceedings of 2011 IEEE International Conference on Computer Vision Workshops (ICCV Workshops), pp. 467-474, 2011.
[9] D. Scharstein and R. Szeliski, "A taxonomy and evaluation of dense two-frame stereo correspondence algorithms," International journal of computer vision, vol. 47, pp. 7-42, 2002.
[10] D. G. Lowe, "Distinctive image features from scale-invariant keypoints," International journal of computer vision, vol. 60, pp. 91-110, 2004.
[11] H. Bay, T. Tuytelaars, and L. Van Gool, "Surf: Speeded up robust features," in Computer Vision–ECCV 2006, ed: Springer, pp. 404-417., 2006.
[12] S. Gauglitz, T. Höllerer, and M. Turk, "Evaluation of Interest Point Detectors and Feature Descriptors for Visual Tracking," International Journal of Computer Vision, vol. 94, pp. 335-360, 2011.
[13] M. Brown and D. G. Lowe, "Unsupervised 3D object recognition and reconstruction in unordered datasets," in Proceedings of Fifth International Conference on 3-D Digital Imaging and Modeling, 3DIM 2005, pp. 56-63, 2005.
[14] S.-Y. Park and M. Subbarao, "A multiview 3D modeling system based on stereo vision techniques," Machine Vision and Applications, vol. 16, pp. 148-156, 2005.
[15] A. Kushal and J. Ponce, Modeling 3d objects from stereo views and recognizing them in photographs: Springer, 2006.
[16] Y. Furukawa and J. Ponce, "Accurate, dense, and robust multiview stereopsis," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 32, pp. 1362-1376, 2010.
[17] http://zh.wikipedia.org/wiki/%E4%B8%89%E7%B6%AD%E6%8E%83%E6%8F%8F%E5%84%80.
[18] R. Hartley and A. Zisserman, Multiple view geometry in computer vision vol. 2: Cambridge Univ Press, 2000.
[19] A. Fusiello, E. Trucco, and A. Verri, "A compact algorithm for rectification of stereo pairs," Machine Vision and Applications, vol. 12, pp. 16-22, 2000.
[20] http://opencv.org/.
[21] C. Harris and M. Stephens, "A combined corner and edge detector," in Alvey vision conference, 1988, p. 50.
[22] E. Rosten and T. Drummond, "Machine learning for high-speed corner detection," in Computer Vision–ECCV 2006, ed: Springer, 2006, pp. 430-443.
[23] P. Viola and M. Jones, "Rapid object detection using a boosted cascade of simple features," in Proceedings of the 2001 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2001, pp. I-511-I-518 vol. 1, 2001.
[24] T. Lindeberg, "Feature detection with automatic scale selection," International journal of computer vision, vol. 30, pp. 79-116, 1998.
[25] H. P. Moravec, "Obstacle avoidance and navigation in the real world by a seeing robot rover," DTIC Document1980.
[26] http://en.wikipedia.org/wiki/Corner_detection.
[27] M. A. Fischler and R. C. Bolles, "Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography," Communications of the ACM, vol. 24, pp. 381-395, 1981.
[28] R. C. Gonzalez and R. E. Woods, "Digital image processing, 2nd," SL: Prentice Hall, 2002.
[29] M. W. Spong, S. Hutchinson, and M. Vidyasagar, Robot modeling and control: John Wiley & Sons New York, 2006.
[30] Z. Zhang, "A flexible new technique for camera calibration," , IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 22, pp. 1330-1334, 2000.
[31] http://francemapping.free.fr/Portfolio/Prog3D/CMVS.html.