臺灣博碩士論文加值系統

本論文主要探討水下動態目標物追蹤與估測的問題，適用於水下複雜環境之即時偵測，能夠有效降低光源、複雜背景與遮蔽物的干擾。本實驗主要是透過影像視覺偵測目標物之後，結合卡爾曼濾波器估測出目標物下一瞬間的位置，達到提升運算速率，有效降低在畫面中相似目標物的背景干擾，使PTZ攝影機能夠自動的追蹤目標物，並框選於畫面中央位置。
在實驗方面，本論文整合影像視覺回授、卡爾曼濾波器估測與PTZ攝影機控制，期望達到水下動態目標物追蹤與估測能力。經實驗結果顯示，卡爾曼濾波器所估測出的估測位置大致都能超前於量測所得到的量測位置；對於強烈光源的影響，則利用色彩空間YCrCb有效的將光源分離。在降低隨機雜訊的部分，結合中值濾波器與型態學中的侵蝕與膨脹，再利用像素相連通處理法處理雜訊，分離出目標物與背景中的相似物。本系統經證實可完成水下目標物追蹤與估測的任務。

The objective of the thesis is to investigate the problems of the dynamic tracking and the states estimation of underwater object. The results of the research can contribute to the real-time detection of underwater object in a complex environment, by reducing the influence of light and the interferences from the background or foreground objects. In the experiment, after the image of target object was captured, we employed Kalman filtering technique to predict object's position at the next instant to enhance the image process speed. Then, by reducing the interferences from the surrounding objects with similar characters, the PTZ camera could automatically track the target by capturing the target in a square window and keeping it close the center area of the screen.
The research integrated the techniques of visual feedback, Kalman filter estimation and controlling of PTZ cameras, to achieve the abilities of dynamic underwater target tracking and estimation. The experimental results showed that the predicted position estimated by the Kalman filter would lead the corresponding measured position approximately. For the influence of a strong light source, we adopted the YCrCb color space to reduce the light effect. Image process techniques such as median filter, erosion and dilation, and connect component were integrated to deal with the influences of random noises, so as to separate the target from the background images. The system was proved to be effective in the underwater target tracking and estimation.

致謝 I
中文摘要 III
ABSTRACT IV
圖目錄 VIII
表目錄 X

第一章 緒論 1
1.1 前言 1
1.2 文獻回顧 5
1.3 研究動機與目的 8
1.4 論文架構 10
第二章 理論背景 11
2.1 數位影像處理(DIGITAL IMAGE PROCESSING) 11
2.1.1 RGB(Red Green Blue)圖像模型 12
2.1.2 色彩模型(Color Space) 13
2.1.3 灰階化（Gray Scale Manipulation） 17
2.1.4 影像二值化（Binarize Image） 18
2.1.5 中值濾波器（Median Filter） 19
2.1.6 型態學-膨脹（Dilation）、侵蝕(Erosion) 20
2.1.7 像素相連通(Connect Component)處理法 23
2.2 卡爾曼理論(KALMAN FILTER) 24
第三章 實驗流程與架構 31
3.1 硬體環境 33
3.1.1 防水PTZ攝影機 34
3.1.2 影像擷取卡 35
3.1.3 水下載具 (ROV) 35
3.2 軟體環境 36
3.2.1 影像擷取 38
3.2.2 PTZ控制系統 38
3.3 卡爾曼濾波器之運算流程 42
3.4 實驗方法 43
第四章 實驗結果與討論 45
4.1 影像擷取參數 45
4.2 實驗規劃與概述 46
實驗1:測試目標物向各個方向移動的狀況 47
實驗2:單純背景下目標物追蹤 49
實驗3:複雜背景環境下無卡爾曼濾波器估測 51
實驗4:複雜背景環境下加入卡爾曼濾波器估測 53
實驗5:目標物移動過遠導致失敗狀況 59
4.3 結果與討論 60
4.3.1 實驗討論與說明 60
實驗1 60
實驗2 60
實驗3 61
實驗4 61
實驗5 62
4.3.2 實驗誤差分析 62
第五章 結論與未來展望 67
5.1 結論 67
5.2 未來展望 68
參考文獻 69

[1]http://www.itri.org.tw/chi/cl/p4.asp?RootNodeId=070&NavRootNodeId=0751&NodeId=075121&ArticleNBR=989
[2]鄧自立編著, ”卡爾曼濾波與維納濾波-現代時間序列分析方法”哈爾濱工業大學出版社, 2003。
[3]J. A. Leese, C. S. Novak, V. R. Taylor, “The Determination of Cloud MotionPatterns from Geosynchronous Satellite Image Data,” Pattern Recognition, Clustering, Statistics, Grammars, Learning, Vol.2, pp. 272-292,1970.
[4]K. P. Horn, and B. G. Schunck, “Determining optical flow,” Artificial Intelligence,Vol.17, pp. 185-203, 1981.
[5]C. S. Fuh, and P. Maragos, “Region-Based Optical Flow Estimation,” Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, San Diego, CA, pp. 130-133, 1989.
[6]Y. P. Tan, S. R. Kulkarni, P. J. Ramadge, “A New Method for Camera Motion Paramater Estimation,” IEEE International Conference on Image Processing, vol. 1, pp. 406-409, 1995.
[7]P. Viola, and J.J. Michael, “Rapid Object Detection Using a Boosted Cascade of Simple features” IEEE CVPR ,Vol.1, NO. 2, pp 511-518, Dec. 2001.
[8]Y. Freund, and R. Schapire, “A Decision-Theoretic Generalization of On-Line Learning and an Application to Boosting”Journal of Computer and System Sciences, 55(1), pp119–139, Aug 1997.
[9]J. Wu, Rehg, J.M., Mullin, M.D., “Learning a Rare Event Detection Cascade by Direct Feature Selection” NIPS, 2003.
[10]D. A. Kulkarni, “Computer Vision and Fuzzy-Neural Systems”, Prentice Hall, Inc., 2001.
[11]D. Marr, and T. Poggio, “Cooperative computation of stereo disparity, ” Science, Vol. 194, pp. 283-287, 1976.
[12]S.T. Barnard and W.B. Thompson, “Disparity analysis of images,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 2, pp. 330-340,1980.
[13]J.S. Lee, C.W. Seo, E.S. Kim, “Implementation of opto-digital stereo object tracking system,” Optics Communications, Vol. 200,pp. 73-85, 2001.
[14]A.P. Tirumalai, B.G. Schunck, R.C. Jain, “Dynamic stereo with self-calibration,” IEEE ransactions on Pattern Analysis and Machine Intelligence, Vol. 14, pp. 1184-1189, 1992.
[15]H. Kase, N. Maru, A. Nishikawa, S. Yamada, and F. Miyazaki, “Visual servoing of the manipulator using the stereo vision,” in Proceedings of the 1993 IEEE/IECON International Conference on Industrial Electronics, Control, and Instrumentation, pp. 1791-1796, 1993.
[16]W. Y. Yau and H. Wang, “Fast relative depth computation for an activestereo vision system,” Real-Time Imaging, Vol. 5, pp. 189-202, 1999.
[17]D. M. Lyons, 2003, “Discrete Event Modeling of Misrecognition in PTZ Tracking,” Proceedings - IEEE International Conference on Advance Video and Signal Based Survillance, 2003.
[18]H. T. Nguyen and W.M. Arnold Smeulders, “Fast Occluded Object Tracking by aRobust Appearance Filter,” Published by the IEEE Computer Society, 2004.
[19]Q. Wan and Y. Wang, “Multiple Moving Objects Tracking under Complex Scenes,” College of Electric and Information Engineering University of Hunan Changsha, Hunan Province, 410082 China. 2006,
[20]C.S. Yang , R.H. Chen and C.Y. Lee and S.J. Lin, “PTZ Camera Based Position Tracking in IP Surveillance System,” 3rd International Conference on Sensing Technology, Nov. 30 – Dec. 3, 2008, Tainan, Taiwan2008.
[21]C.P. Papageorgiou, M. Oren, and T. Poggio, “A General Framework for Object Detection”In nternational Conference on Computer Vi-sion ,pp555-562 Jan. 1998.
[22]A. D. Francesca, O. A. Verri, “A trainable system for face detection in unconstrained environments,” 14th International Conference on Image Analysis and Processing , 2007,.
[23]S. Huttunen and J. Heikkil‥, “An Active Head Tracking System for Distance Education and Videoconferencing Applications,” Proceedings of the IEEE International Conference on Video and Signal Based Surveillance IEEE , 2000.
[24]G. S. Hornby, S. Takamura, T. Yamamoto and M. Fujita, “Autonomous Evolution of Dynamic Gaits With Two Quadruped Robots,” IEEE Transactions on Robotics, Vol. 21, No. 3, pp. 402-410, 2005.
[25]D Jang, H. Choi, ”Moving Object Tracking Using Active Models,” Image Processing, vol.3,pp.648-652,Oct. 1998.
[26]L. J. Latecki, R. Miezianko, “Object Tracking with Dynamic Template Update and Occlusion Detection,” Proceedings of the 18th International Conference on Pattern Recognition IEEE, 2006.
[27]B.K. Horn, J.E. Schunck, “Determining Optical Flow” Artificial Intelligence, Vol 17, pp185-203, 1981.
[28]C. Chang, R. Ansari, A. Khokhar, “Multiple Object Tracking with Kernel Particle Filter,” Proceedings of the 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2005.
[29]W. H. Li and L. Kleeman, “Real Time Object Tracking using Reflectional Symmetry and Motion,” Proceedings of the 2006 IEEE /RSJ International Conference on Intelligent Robots and Systems October 9-15, 2006, Beijing, China.
[30]R. Lin, Z. Du, F. He, M. Kong , L. Sun, ”Tracking a moving object with mobile robot based on vision”, International Joint Conference on Neural Networks,2008.
[31]P. Goolkasian, “Processing Visual-Stimuli Inside and Outside the Focus Fattention,” Bulletin of the Psychonomic Society, vol. 29, no. 6, pp.510-515, 1991.
[32]陳偉銘、趙涵捷編著，2001，影像裡的數學世界，台灣書店。
[33]繆紹綱編著，數位影像處理活用Matlab，全華科技，2004。
[34]http://www.cppblog.com/Amigo/archive/2008/04/01/45914.html ,作者：Jearome
[35]張家鳴, 利用視覺輔助力量回授於物件輪廓掃描,國立台灣海洋大學機械與機電工程研究所碩士論文, 2006。
[36]吳承勳, 改良式多重目標物之追蹤演算法,國立台灣海洋大學電機工程研究所碩士論文, 2008。