中文摘要

隨著機械與資電科技的進步，電腦視覺已成為一項熱門的研究領域，凡舉日常生活的視訊會議與防盜監視、工業生產線的自動化視覺檢測與條碼判讀等都已成為應用中不可缺少的一部份。電腦視覺系統特別能勝任繁複而需耗人力來全神灌注的工作上，對於追蹤、攔截、監督、檢測、定位與紀錄等工作特別有效率。
本研究主要工作在建立一模仿人類之視覺系統與手臂運動的配合，進而完成三維空間中物件攔截之任務。研究包含兩個重要工作，第一部份為機械視覺系統，也就是以電腦與雙眼機械頭來模擬人的頭部及雙眼的運動並連結思考決策的大腦。視覺系統主要工作為數位影像處理，與採用一快速強健的橢圓追蹤演算法來追蹤影像中的特徵物體。雙攝影機經過校準後可精確的重建動態三維定位資訊並具有錯誤鎖定的判斷能力。第二部分為運動控制系統，可看成人類的肌肉與神經系統，電腦在中斷程式中即時控制五軸，負責操控整個三軸機械頭部的影像追蹤運動與二軸機械手臂對物體的攔截運動。在影像目標估測中，本研究應用灰色估測模型來補償CCD取像時產生的量化誤差。在雙眼機械頭影像追蹤方面，使用以「視覺為基礎」的視覺伺服控制，誤差訊號定義在影像平面上來完成追蹤目標的運動；而在機械手臂攔截方面，則是使用以「位置座標為基礎」的視覺伺服控制，左右攝影機目標物位置經由攝影機模型求得空間座標，誤差訊號定義在此空間座標中，而兩控制法則的控制訊號皆直接送入運動控制器中，然後經由驅動器驅動馬達，完成運動控制。本系統之雙眼機械頭能緊追落於視覺範圍內之黑色球狀目標物，進而完成三維空間中物件攔截之任務。

Abstract

In recent years, with the excellent progress in mechantronics and computer science, computer vision had become an active research field. It also has many applications, such as video conference in daily life, automatic industrial visual inspection, matrix code reading in product line and the image tracking system in military and national defense, etc… and its impact becomes more important than before. Especially, it has great competence and efficiency for heavy and repeated jobs, that are time consuming and a lot of human efforts are required.
In this study, a robot system has been used to perform a real time "watching and catching" task. It consists of two major parts. The first is the vision system, which handles the image processing and image understanding and provides useful information for tracking task. The second is the motion control system, which contains software and hardware of the robot and the "head-eye" system. The vision system deals with real-time digital image processing with a high-speed, robust elliptical tracking algorithm to track object features. The mechanism of the head eye system can reconstruct precise 3D configuration with ability of error detection and active model validation with CCD camera calibration.
The real time "watching and catching" task is carried by position based look and move control structure, where the tracking error signal is defined by the spatial pose of the target with respect to the robot.
The gray modeling has also been applied to filter the target position and produce reliable information of the target. The results shown in this thesis has demonstrated a great potential of real time visual servo to industrial application.

參考文獻
[1] A.C. Sanderson and L.E. Weiss, "Image-based visual servo control using relational graph error signals", Proc. IEEE, Page(s): 1074-1077, 1980.
[2] A.M. Baumberg and D.C. Hogg,. "An efficient method for contour tracking using active shape models, Motion of Non-Rigid and Articulated Objects", Proceedings of the 1994 IEEE Workshop, Page(s): 194-199, 1994.
[3] B.G. Schunck and B.K.P. Horn, "Constraints on optical flow computation", Proc. Pattern Recognition and Image Processing Conf., Dallas, Page(s): 205-210, 1981.
[4] C.H. Lin and J.L. Wu, "Automatic facial feature extraction by genetic algorithms", IEEE Transactions on Image Processing, Volume 86, Page(s): 834-845, June 1999.
[5] C. Wiles and A. Davison, "Multi-View Modeling and Analysis of Visual Scenes", IEEE Workshop, 1999, (MVIEW '99) Proceedings, Page(s): 29-36, 1999.
[6] D.V. Papadimitriou and T.J. Dennis, "Epipolar line estimation and rectification for stereo image pairs", IEEE Transactions on Image Processing, vol.5, no.4, Page(s): 672-676, April 1996.
[7] E. Marchand, "VISP: a software environment for eye-in-hand visual servoing", Proceedings of IEEE International Conference on Robotics and Automation, Detroit, Michigan, Vol.4, Page(s): 3224-3229, 1999.

[8] F. Li, M. Brady and C. Wiles, "Fast computation of the fundamental matrix for an active stereo vision system", Computer Vision, In Procceding of ICCV'96 International Conference on Page(s): 157-166, 1996.
[9] G. Wei and S. Ma, "Implicit and explicit camera calibration: Theory and experiments", IEEE Transactions on Pattern Analysis and Machine Intelligence, 16(5) Page(s): 469-840, 1994.
[10] G. Wei, K. Arbter. and G. Hizinger, "Active Self Calibration of Robotic Eyes and Hand-Eye Relationships with Model Identification", IEEE Trans. On Robotics and Automation, Vol. 14 no.1 February, 1998.
[11] H. Kass, A. Witkin and D. Terzopoulos, "Snakes: Active contour models", International Journal of Computer Vision. Vol. 31, Page(s): 1-321, 1987.
[12] H. Liu, R.C. Ramon and L.S. Paulo, "Stable Adaptive Visual Servoing for Moving Targets" Proceedings of the American Control Conference, Chicago, Illinois, Page(s): 2008-2012, 2000.
[13] J. Luettin, N.A. Thacker, S.W. Beet, "Locating and tracking facial speech features", Pattern Recognition, 996, Proceedings of the 13th International Conference on Volume: 1, Page(s): 652-656, 1996.
[14] Levenberg Marquardt Algorithm Source Code
URL:http://www-fp.mcs.anl.gov/otc/Guide/OptWeb/continuous/unconstrained/nonlinearls/section2_1_2.html
URL:http://www.ifor.math.ethz.ch/~finschi/Papers/LevMar.en.html
[15] M. Isard and A. Blake. "The Condensation Algorithm",
URL:http://www.robots.ox.ac.uk
[16] M. Li, "Kinematic Calibration of an Active Head-Eye System", IEEE Transactions on Robotics and Automation, vol.14, No.1, February, 1998.
[17] M. Sonka, V. Hlavac and R. Boyle, "Image Processing, Analysis, and Machine Vision", second edition, PWS, Pacific Grove, California, 1998.
[18] N.K. William and A.C. Bovik, "FOVEA: A Foveated vergent active stereo vision system for dynamic three-dimensional scene recovery", IEEE Transactions on robotics and automation, vol.14, no.5, October 1998.
[19] Q. Chen, H. Wu, T. Fukumoto and M. Yachida, "3D Head Pose Estimation without Feature Tracking, Automatic Face and Gesture Recognition", Proceedings in Third IEEE International Conference, Page(s): 88-93, 1998.
[20] R. Tsai, "A Versatile Camera Calibration Technique for High-Accuracy 3D Machine Vision Metrology Using Off-the-Shelf TV Cameras and Lenses", IEEE Trans. Robot. Automat., vol.3, Page(s): 323-344, Aug, 1987.
[21] R. Tsai's camera calibration technique source code
URL:http://www-2.cs.cmu.edu/afs/cs.cmu.edu/user/rgw/www/TsaiCode.html
[22] R. Klette and A. Koschan, "Computer Vision, Three-Dimensional Data from Images", Springer, Singapore, 1998.
[23] S. Birchfield, "Elliptical Head Tracking Using Intensity Gradients and Color Histograms", In IEEE Conf. on Computer Vision and Pattern Recognitions, Santa Barara, California, June 1998.
[24] S. Birchfield, "An elliptical head tracker", In Proc. Of the 31st Asilomar Conf. On Signal, Systems and Computers, 1997.
[25] S. Hutchison, G.D. Hager and P.I. Corke, "A Tutorial on Visual Servo Control", IEEE Transaction on Robotics and Automation, vol.12, No.5, Page(s): 651-668, 1996.
[26] V. Bakic and G. Stockman, "Real-time tracking of face features and gaze direction determination", Fourth IEEE Workshop, Applications of Computer Vision, 1998. WACV'98. Proceedings, Page(s): 256-257, 1998.
[27] W.Y. Yau and H. Wang, "Coordinating the eyes, head and arm of an autonomous robot", Engineering Applications of Artificial Intelligence 11, Page(s): 163-174, 1998.
[28] X. Xie, R. Sudhakar and H. Zhuang, "Real-time eye feature tracking from a vedio image sequence using Kalman filter", IEEE Transactions on Systems, Man and Cybernetics, Volume: 25, Page(s): 1568-1577, 12, Dec. 1995.
[29] Y. Matsumoto and A. Zelinsky. "Real-time Stereo Face Tracking for Visual Human Interfaces, Recognition, Analysis, and Tracking of Faces and Gestures in Real-Time Systems", International Workshop Proceedings, Page(s): 77-82, 1999.
[30] Z. Zhang, "A Flexible New Technique for Camera Calibration", Microsoft Research, Technical Report MSR-TR-98-71, March, 1999.
[31] 李尚昇，"影像理解與決策在對奕程序之應用"，國立成功大學碩士論文，1999
[32] 吳承柯，戴善榮，程湘君，雲立實，"數位影像處理"，儒林，1993
[33] 林柏勳，"利用影像技術於測定機械臂二維工作空間座標之研究"，國立成功大學碩士論文，2000。
[34] 林思宏，"灰色建模在預測控制上之應用"，國立成功大學碩士論文，1992
[35] 林振輝，"視覺迴授在3D軌跡追蹤之應用"，國立成功大學碩士論文，1997
[36] 黃文明，"視覺迴授伺服控制系統之研究"，國立成功大學碩士論文，1996
[37] 黃細田，"在多工環境下利用機械視覺於3D軌跡重建之研究"，國立成功大學碩士論文，2001
[38] 謝鋒儒，"機器人頭與手臂之協調控制"，國立成功大學碩士論文，2000
[39] 譚永恆，"以數位影像處理技術做人臉自動追蹤系統之研究"，國立成功大學碩士論文，2000