傳統的單槓體操選手訓練過程依賴教練肉眼觀察選手的動作表現，然而，這種方法無法精確量化和評估選手的動作參數，如滯空高度、滯空時間和旋轉速度等。因此，選手無法全面了解自己的動作表現，進而可能限制了他們能力的提升。為了解決這些限制，本研究旨在開發一種基於影像處理的單槓動作參數量測系統，以提供更精確和全面的動作參數量化評估，可以免除在選手身上安裝穿戴式裝置的問題，也較易於攜帶。
本論文透過背景分割(Background subtraction)的方式能夠完整的提取體操選手的人物輪廓，以及霍夫轉換(Hough transformation)來對單槓兩側鋼索進行直線檢測，進而對單槓進行定位，通過提取選手輪廓，以及單槓位置後，再透過本論文所開發之動作參數分析演算法來分析選手的滯空高度、滯空時間以及旋轉速度等等動作表現。
本論文針對不同單槓動作中的飛行高度進行量測，如貓跳(Yamawaki)和特卡契夫(Tkatchev)，也針對大迴環(Giant swing)之旋轉速度進行量測。我們對4位受試者共47筆資料進行分析與量測。在滯空高度量測部分，貓跳的平均絕對誤差(MAE)和標準差(SD)分別為3.10公分和2.81公分，特卡契夫的平均絕對誤差和標準差分別為2.01公分和1.28公分。在滯空時間量測部分，貓跳的平均絕對誤差和標準差分別為0.029秒和0.023秒，特卡契夫的平均絕對誤差和標準差分別為0.035秒和0.029秒。在旋轉速度量測部分，大迴環的平均絕對誤差和標準差分別為5.657度/秒以及1.117度/秒。在飛行次數和旋轉圈數計算則是高達100 %的準確度。

Traditional training methods for horizontal bar gymnasts rely on coaches' visual observation of athletes' performance. However, this approach lacks of precise quantifying and evaluating key motion parameters, such as flight height, flight time and rotation speed. This limitation may hinder athletes' ability to fully understand their performance, potentially impeding their progress. To address these limitations, this thesis aims to develop an image processing-based system for accurately quantifying and evaluating gymnasts' motion parameters, providing a comprehensive assessment without the need for wearable devices on athletes and ensuring ease of portability.
In this thesis, the proposed system employs background subtraction to extract the gymnast's silhouette from the video, as well as Hough transformation to detect the straight lines of the horizontal bars cables. This allows the system to locate the position of the horizontal bars accurately. After obtaining the gymnast's silhouette and position of the horizontal bar, the developed algorithm analyzes the motion parameters, such as flight height, flight time, and rotation speed.
This thesis focuses on the measurement of flight height in different horizontal bar exercises, such as the Yamawaki and Tkatchev, as well as the measurement of rotational speed in the Giant swing. We evaluated 47 data from 4 participants. In the part of flight height measurement, the mean absolute error (MAE) and standard deviation (SD) for Yamawaki were 3.10 cm and 2.81 cm, respectively, while for Tkatchev, they were 2.01 cm and 1.28 cm, respectively. In the part of flight time measurement, the mean absolute error and standard deviation for Yamawaki were 0.029 seconds and 0.023 seconds, respectively, while for Tkatchev, they were 0.035 seconds and 0.029 seconds, respectively. Regarding the measurement of rotational speed, the mean absolute error and standard deviation were 5.657 degrees/second and 1.117 degrees/second, respectively. And the calculations for number of flight and rotations achieved 100 % accuracy.

[1] Chinese Taipei Olympic Committee, “Tang Chia-Hung,” https://www.tpenoc.net/athlete/chia-hung-tang/.(accessed April 17, 2023).
[2] Federation International Gymnastics. “TANG Chia-Hung,” https://www.gymnastics.sport/site/athletes/bio_detail.php?id=36431.(accessed June. 2, 2023).
[3] I. Čuk, A. Atiković, and M. Tabaković, “Tkachev salto on high bar,” Science of Gymnastics Journal, vol. 1, no. 1, pp. 5-13, Jan. 2009.
[4] H. Tsai and W. J. Wang, “Analysis native athletes' Tkatchov skill,” Bulletin of Physical Education, National Society of Physical Education, R.O.C., vol. 28, pp. 293-304, Mar. 2000, doi:10.6222/pej.0028.200003.4029.
[5] K. Spencer and M. Schuhmann, “The influence of body position on the straddled Tkatchev’s flight phase in men’s horizontal bar,” Journal of Human Sport and Exercise, vol. 12, no. 1, pp. 204-218, May. 2017, doi:10.14198/jhse.2017.121.17.
[6] A. Arampatzis and G.P. Brüggemann, “Mechanical energetic processes during the giant swing before the Tkatchev exercise,” Journal of Biomechanics, vol. 34, no. 4, Apr. 2001, doi: 10.1016/s0021-9290(00)00212-8. PMID: 11266674.
[7] M. J. Hiley and M. R. Yeadon, “Maximal dismounts from high bar,” Journal of Biomechanics, vol. 38, no. 11, pp. 2221-2227, Nov. 2005, doi: 10.1016/j.jbiomech.2004.09.025.
[8] M. J. Hiley, V. V. Zuevsky, and M. R. Yeadon, “Is skilled technique characterized by high or low variability? An analysis of high bar giant circles,” Human Movement Science, vol. 32, no. 1, pp. 171-180, Feb. 2013. doi: 10.1016/j.humov.2012.11.007.
[9] N. Arakawa, K. Ohtsuka, and A. Yoshio, “Investigation of the athlete’s motion using the gymnastics apparatus’s motion,” Proceedings of the 13th Conference of the International Sports Engineering Association, vol. 49, no. 1, pp. 120, June. 2020, doi:10.3390/proceedings2020049120.
[10] Y. Zhao et al., “Quantitative evaluation of gymnastics based on multiple MEMS sensors,” in IEEE Sensors Journal, vol. 21, no. 21, pp. 24531-24539, 1 Nov.1. 2021, doi: 10.1109/JSEN.2021.3114758.
[11] B. Chen, L. Kuang, and W. He, “Simulation of gymnastics performance based on MEMS sensor,” EURASIP Journal on Advances in Signal Processing, vol. 2021, no. 1, pp. 117, Jul. 2021, Art. no. 47, doi:10.1186/s13634-021-00760-4.
[12] P. Li and J. Zhou, “Tracking of gymnast’s limb movement trajectory based on MEMS inertial sensor,” Applied Bionics and Biomechanics, vol. 2022, Apr. 2022, Art. no. 5292454, doi:10.1155/2022/5292454.v
[13] M. P. Díaz-Pereira, I. Gómez-Conde, M. Escalona, and D. N. Olivieri, “Automatic recognition and scoring of olympic rhythmic gymnastic movements,” vol. 34, pp. 63-80, Apr. 2014, doi: 10.1016/j.humov.2014.01.001.
[14] J. Omorczyk, L. Nosiadek, T. Ambroży, and A. Nosiadek, “High-frequency video capture and a computer program with frame-by-frame angle determination functionality as tools that support judging in artistic gymnastics”, Acta of Bioengineering and Biomechanics, vol. 17, no. 3, pp. 85-93, Jan. 2015, doi: 10.5277/ABB-00123-2014-02.
[15] H. Shimamoto, T. Taki, and J. Hasegawa, “A study of an automated scoring system for the twist skill in horizontal bar of artistic gymnastics”, ECTI Transactions on Computer and Information Technology, vol. 7, no. 1, pp. 1–6, May. 2013, doi: 10.37936/ecti-cit.201371.54345.
[16] B. Reily, H. Zhang, and W. Hoff, “Real-time gymnast detection and performance analysis with a portable 3D camera”. Computer Vision and Image Understanding, vol. 159, pp. 154-163, Nov. 2016, doi: doi.org/10.1016/j.cviu.2016.11.006.
[17] C. Schärer et al., “Simple assessment of height and length of flight in complex gymnastic skills: validity and reliability of a two-dimensional video analysis method”, Applied Sciences, vol. 9, no. 19, pp. 3975, Oct. 2019, doi:10.3390/app9193975.
[18] 中華奧林匹克委員會. “競技體操”, https://www.tpenoc.net/sport/artistic-gymnastics/.(accessed Feb.1, 2023).
[19] Y. Wu, J. Lim, and M.H. Yang, “Online object tracking: a benchmark,” CVPR '13: Proceedings of the 2013 IEEE Conference on Computer Vision and Pattern Recognition, pp. 2411-2418, June. 2013, doi: 10.1109/CVPR.2013.312.
[20] OpenCV (4.7.0-dev), “How to use background subtraction methods,” https://docs.opencv.org/4.x/d1/dc5/tutorial_background_subtraction.html.(accessed Mar.14, 2023).
[21] Z. Zivkovic, “Improved adaptive Gaussian mixture model for background subtraction,” Proceedings of the 17th International Conference on Pattern Recognition, Vol. 2, pp. 28-31, doi: 10.1109/ICPR.2004.1333992.
[22] Z. Zivkovic and F. van der Heijden, “Efficient adaptive density estimation per image pixel for the task of background subtraction,” Pattern Recognition Letters, vol. 27, no. 7, pp. 773-780, May. 2006. doi: 10.1016/j.patrec.2005.11.005.
[23] J. Serra, “Image analysis and mathematical morphology,” New York, NY, 1982, doi:10.1002/cyto.990040213.
[24] R. M. Haralick, S. R. Sternberg, and X. Zhuang, "Image analysis using mathematical morphology", IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. PAMI-9, no. 4, pp. 532-550, Jul. 1987. doi: 10.1109/TPAMI.1987.4767941.
[25] P. V. C. Hough, “Method and means for recognizing complex patterns,” U.S. Patent 30696541962.
[26] R. O. Duda and P. E. Hart, “Use of the Hough transformation to detect lines and curves in pictures,” Communications of the ACM, vol. 15, no. 1, pp. 11-15, Jan. 1972, doi:10.1145/361237.361242.
[27] OpenCV (4.7.0-dev), “Hough Line Transform,” https://docs.opencv.org/4.x/d1/dc5/tutorial_background_subtraction.html.(accessed July.24, 2023).
[28] OpenCV (4.7.0-dev), “Color Space Conversions,” https://docs.opencv.org/4.7.0/d8/d01/group__imgproc__color__conversions.html.(accessed Aug.20, 2023).
[29] OpenCV (4.7.0-dev), “Smoothing Images,” https://docs.opencv.org/4.7.0/d4/d13/tutorial_py_filtering.html.(accessed July.24, 2023).
[30] J. Canny, “A computational approach to edge detection”, in IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. PAMI-8, no. 6, pp. 679-698, Nov. 1986, doi: 10.1109/TPAMI.1986.4767851.
[31] OpenCV (4.7.0-dev), “Contour Features,” https://docs.opencv.org/4.7.0/dd/d49/tutorial_py_contour_features.html.(accessed July.17, 2023).
[32] 中華民國體操協會, “競技體操規則,” https://www.ctga.com.tw/WordPress/wp-content/uploads/2019/10/%E7%94%B7%E5%AD%90%E9%AB%94%E6%93%8D%E8%A6%8F%E5%89%87.pdf.(accessed Mar.19, 2023).
[33] 30 Minutes of everything, “where is your center of gravity?,” https://30minutesofeverything.com/where-is-your-center-of-gravity/.(accessed July.24,2023).
[34] J. Tsuchiya, K. Murata, and T. Fukunaga, “Kinetic analysis of backward giant swing on parallel bars,” International Journal of Sport and Health Science, vol. 2, pp. 211-221, Aug. 2004, doi: 10.5432/ijshs.2.211.
[35] J. Charmant and contributors. (2021) “Kinovea (Version 0.9.5),” https://www.kinovea.org.(accessed Aug.6, 2023).