本論文以實際應用於RoboCup世界盃雙足人形機器人競賽為目的，提出一基於卡爾曼濾波器之中型雙足人形機器人足球影像追蹤與定位系統，達到自主足球追蹤與定位之目的。往年RoboCup競賽之競賽場地皆為綠色地毯，場地上分佈白色場地線，且雙足人形機器人競賽使用橘色足球與黃色球門。因此，在往年競賽中，參賽隊伍大多使用顏色作為主要辨識特徵來偵測目標物。然而，與往年不同的是賽事主辦單位調整今年之競賽規則，將足球與球門均更改為白色，使得在球場上，足球主要顏色、場地線與球門都是白色，這對於傳統以顏色特徵作為目標物偵測方法將無法於今年競賽中使用，大幅提升了機器人對於目標物影像偵測之困難度。為了克服上述問題，本論文提出一針對HSV色彩模型進行線性迴歸分析之方法，以迴歸分析建立一更精準之色彩區間，達到更精確地過濾綠色場地以利偵測場地上之線段及物體，再從場地上之線段及多重物件中找到足球。同時使用卡爾曼濾波器進行足球移動路徑預估，由預估球之移動位置來建立一動態影像子視窗；此一動態影像子視窗之尺寸為根據預估球在影像中之尺寸乘上一由追蹤誤差定義出之公差值所產生。透過針對子視窗之畫面進行影像處理，達到降低背景環境干擾並且提升影像處理速度，達到更準確有效之偵測與追蹤結果。此外透過場地影像定位系統可將足球之影像座標轉換成相對於機器人之場地座標。最後，此一系統所提出之方法在與非使用卡爾曼濾波器之實驗結果比較下，影像計算時間降低了49.03%。最後將此系統實際應用於2015 RoboCup世界盃機器人競賽之中，並在該競賽中取得了足球競賽項目第二名，此一結果證明本系統之可行性。

This thesis proposes a Kalman filter (KF) based visual soccer ball tracking and localization approach for an autonomous teen-size biped humanoid robot. This approach is capable of dealing with the visual tracking function in the humanoid league of RoboCup 2015. The competition field was covered by a piece of green carpet painted with white lines in former RoboCup humanoid leagues. Moreover, two identical yellow soccer goals and an orange color ball were used. Therefore, color-based ball recognition is practically used in formers competitions because of uses of distinct colors of ball, goal and field lines. Nevertheless, RoboCup 2015 announced a significant alteration in the specification of ball and goal colors. The major colors of the ball and two goals have been altered as a white color which is the same as the field line. It would be a great challenge to the color recognition when compared to former events, and the approaches with only applying color attribute thresholds. To overcome the aforementioned challenge, this work initially filters out the green carpet color with a linear regressive color space model to precisely retrieve all foreground objects such as robots, ball, goals and filed lines. Subsequently, the soccer ball is identified from the retrieved foreground objects. Especially, a KF is used to predict the ball moving trajectory. The predicted ball moving trajectory is further used to identify the ball position of the next image sampling time. According to the predicted ball position, a dynamic region of interest (ROI) can be generated. Its adjustable ROI dimension is formed with respect to the predicted ball image size with multiplying a tolerance factor which depends on the tracking error range. The dynamic ROI is beneficial to significantly reduce the interference of other objects in the field as well as to improve the computational efforts. Furthermore, a floor localization approach is used to transform the pixel coordinate system to the floor coordinate system. Finally, all proposed approaches have been realized and evaluated in this paper. When compared to the system without employing KF; the computational time of ball tracking is reduced 49.03%. The proposed algorithm has keep applied to our humanoid teen size robot, HuroEvolution TN, and it participated in the soccer game of RoboCup 2015 to practically evaluate the performance. As a consequence, the HuroEvolution TN awarded the second place. The results also verified the feasibility of the proposed approach.

[1] Ganesan P, V.Rajini, B.S. Sathish, Khamar Basha.Shaik, “The application of improved HSV color space model in image processing,” International Conference on Control, Instrumentation, Communication and Computational Technologies, pp. 101 – 105, 2014.
[2] Tse-Wei Chen, Yi-Ling Chen, Shao-Yi Chien, “Fast Image Segmentation Based on K-Means Clustering with Histograms in HSV Color Space,” IEEE Workshop on Multimedia Signal Processing, pp. 322 – 325, 2008.
[3] Li Shuhua, Guo Gaizhi, “HSV Color Space Based Segmentation of Region of Interest in Satellite Images,” International Conference on Future Computer and Communication, vol. 2, pp. 10 – 13, 2010.
[4] 盧俊良，「基於光線與臉部表情變化下之人臉辨識」，碩士學位論文，國立中央大學，民國98年。
[5] Hassan Fahmy Hashem, “Adaptive Technique for Human Face Detection Using HSV Color Space and Neural Networks,” National Radio Science Conference, Faculty of Engineering, Future Univ., Egypt, 2009.
[6] Krissada Asavaskulkeit, Somchai Jitapunkul, “The Color Face Hallucination with the Linear Regression Model and MPCA in HSV Space,” International Conference on Systems, Signals and Image Processing, pp. 1 – 4, 2009.
[7] Jong-Min Jeong, Tae-Sung Yoon, and Jin-Bae Park, “Kalman Filter Based Multiple Objects Detection-Tracking Algorithm Robust to Occlusion,” SICE Annual Conference, Hokkaido University, Sapporo, Japan, pp. 941 – 946, 2014.
[8] Jesse Scott, Michael A. Pusateri, Ph. D., Naohiko Sugita, Duane Cornish, “Kalman Filter Based Video Background Estimation,” Penn State University University Park, Pennsylvania. 14-16, 2009.
[9] Shengluan Huang, Jingxin Hong, “Moving Object Tracking System Based On Camshift And Kalman Filter,” International Conference on Consumer Electronics, Communications and Networks, pp. 1423 – 1426, 2011.
[10] Shuai Zhang, Chong Wang, Shing-Chow Chan, Xiguang Wei, Check-Hei Ho, “New Object Detection, Tracking, and Recognition Approaches for Video Surveillance Over Camera Network,” IEEE Sensors Journal, vol. 15, pp. 2679 – 2691, 2014.
[11] Macharla Vinaykumar, Ravi Kumar Jatoth, “Performance Evaluation of Alpha-Beta and Kalman Filter for Object Tracking,” International Conference on Advanced Communication Control and Computing Technologies, pp. 1369 – 1373, 2014.
[12] Jun Gui, Tao Zhang, Lijun Liu, “The application of Kalman filtering in the system of image difference moving objects tracking,” International Conference on Computational and Information Sciences, pp. 83 – 86, 2011.
[13] Young Min Baek, Shinichi Tanaka, Kanako Harada, Naohiko Sugita, Akio Morita, Shigeo Sora, and Mamoru Mitsuishi, “Robust Visual Tracking of Robotic Forceps Under a Microscope Using Kinematic Data Fusion,” IEEE/ASME Transactions on Mechatronics, vol. 19, pp. 278 – 288, 2012.
[14] Hongxia Chu, Kejun Wang, “Target Tracking Based on Mean Shift and Improved Kalman Filtering Algorithm,” International Conference on Automation and Logistics, pp. 808 – 812, 2009.
[15] S. Y. Chen, “Kalman Filter for Robot Vision: A Survey,” IEEE Transactions on Industrial Electronics, pp. 4409 – 4420, 2011.
[16] Zhonghai Deng, Arjan Gijsenij, Jingyuan Zhang, “Source Camera Identification Using Auto-White Balance Approximation,” IEEE International Conference on Computer Vision, pp. 57 – 64, 2011.
[17] Xiaoqin Zhang, Weiming Hu, Zixiang Zhao, Yan-guo Wang, Xi Li, Qingdi Wei, “SVD Based Kalman Particle Filter for Robust Visual Tracking,” International Conference on Pattern Recognition, pp. 1 – 4, 2008.
[18] Wonsang Hwang, Jaehong Park, Hyun-il Kwon, Anjum, M.L., Jong-hyeon Kim, Changhun Lee, Kwang-soo Kim, Dong-il Dan Cho, “Vision Tracking System For Mobile Robots Using Two Kalman Filters and a Slip Detector,” International Conference on Control Automation and Systems, pp. 2041 – 2046, 2010.
[19] 張國澤，「以卡爾曼濾波器為基礎之運動中錐形單擺球體的定位追蹤」，碩士學位論文，台灣科技大學，民國95年。
[20] Simon Thompson, Satoshi Kagami, “Humanoid Robot Localisation using Stereo Vision,” IEEE-RAS International Conference on Humanoid Robots, pp. 19 - 25, 2005.
[21] 李庚諺，「影像定位與同步建圖技術於導航系統之應用」，碩士學位論文，國立成功大學，民國101年。
[22] 林姿吟，「基於立體視覺的3D影像定位」，碩士學位論文，台灣科技大學，民國100年。
[23] 邱治華，「結合尺度不變特徵與貝氏定理的影像定位與導航系統設計」，碩士學位論文，朝陽科技大學，民國97年。
[24] Iwan Awaludin, Priyanto Hidayatullah, Jonner Hutahaean, Dewa Gede Parta, “Detection and Object Position Measurement using Computer Vision on Humanoid Soccer,” Computer Engineering and Informatics Department, Bandung State Polytechnic, Bandung, Indonesia, 2013.
[25] Muhammad Nuruddin Sudin, Mohammad Faidzul Nasrudin, Siti Norul Huda Sheikh Abdullah, “Humanoid Localisation in a Robot Soccer Competition Using a Single Camera,” IEEE International Colloquium on Signal Processing & its Applications, pp. 77 - 81, 2014.
[26] Hideaki Minakata, Yasuo Hayashibara, Katsuhiro Ichizawa, Takehito Horiuchi, Masahiro Fukuta, Shohei Fujita, Hiroshi Kaminaga, Kiyoshi Irie, Hajime Sakamoto, “A Method of Single Camera Robocup Humanoid Robot Localization using Cooperation with Walking Control,” IEEE International Workshop on Advanced Motion Control, pp. 50 - 55, 2008.
[27] Jen-Shiun Chiang, Chih-Hsien Hsia, Shih-Hung Chang, Wei-Hsuan Chang, Hung-Wei Hsu, Yi-Che Tai, Chun-Yi Li, and Meng-Hsuan Ho, “An Efficient Object Recognition and Self-Localization System for Humanoid Soccer Robot,” SICE Annual Conference, pp. 2269 - 2278, 2010.
[28] Sajjad Torabian, Soroush HoseinAlipour, Alireza Mirzargar, Mehran Tavakkolian, “Improving the Localization of Humanoid Soccer Robots in Specified Fields: A Neural Network Approach,” RSI/ISM International Conference on Robotics and Mechatronics, pp. 443 - 448, 2013.
[29] J.M. Ibarra Zannatha, R. Cisneros Lim´on, A.D. G´omez S´anchez, E. Hern´andez Castillo, L.E. Figueroa Medina, F.J.K. Lara Leyva, “Monocular visual self-localization for humanoid soccer robots,” Departament of Automatic Control, Cinvestav Departament of Computer Systems, Cinvestav Engineering Faculty, La Salle University, pp. 100-107, 2011.
[30] Wei Hong, Changjiu Zhou, Yantao Tian, “Robust Monte Carlo Localization For Humanoid Soccer Robot,” IEEE/ASME International Conference on Advanced Intelligent Mechatronics, pp. 934 - 939, 2009.
[31] Bo Tian, Chuen-Leong Ng, Chee-Meng Chew, “Self-Localization of Humanoid Robots with Fish-eye Lens in a Soccer Field,” IEEE Conference on Robotics, Automation and Mechatronics, pp. 522 – 527, 2010.
[32] Giuseppe Oriola, Antonio Paolillo, Lorenzo Rosa, Marilena Vendittelli, “Vision-Based Odometric Localization for Humanoids using a Kinematic EKF,” IEEE-RAS International Conference on Humanoid Robots, pp. 153 – 158, 2012.
[33] Steven J. Miller, “The Method of Least Squares,” Mathematics Department Brown University, Available at: http://web.williams.edu/Mathematics/sjmiller/public_html/BrownClasses/54/handouts/MethodLeastSquares.pdf .(Access: 2015/08/28)