現今社會下，運動意識越來越普及，更多人會自行選擇去健身房、去附近公園又或者在家中運動。在沒有專業教練指導下，運動姿勢的錯誤，很可能緩慢的對人體造成永久性傷害，故本論文設計一個輔助系統代替專業教練指導，利用語音給予使用者即時回饋，以幫助使用者做出正確的運動姿勢。
本系統之運動姿勢即時語音檢測系採用的硬體有三視角的深度攝影機，配合骨架的抓取，以分析運動姿勢。系統分別給予專業教練與使用者不同的模式。教練模式提供教練錄製教學影像，與標記出需活動的關節；使用者模式則提供使用者運動時，系統即時與教練的資料進行比對，並給予矯正的語音。
本系統資料的比對方法，是將各個運動部位的角度和部分關節與相鄰部位的夾角，與教練的模板中相對應的角度，透過閥值來辨別其姿勢的正確性。同時在運動中記錄其運動部位角度變化，以計算使用者運動次數與動作的完成度，並給予語音建議。而語音建議播放前須通過本系統設計的決策樹，以判別其錯誤部位是否為其他身體部位所影響，來達到最正確的錯誤矯正。
本系統每次同時計算三個視角的影格需時約25.3(ms)，而一個影格時間為33(ms)，有達到即時的需求，其整體正確率達到96.1%。經過實際測試，已可以供使用者順利使用，且專業教練可以隨時補充運動項目。

Nowadays, the sports consciousness is gradually raising. More people spend their time exercising in gymnasium, in park, or at home. However, exercising with wrong posture may cause permanent harm to our body. Therefore, this study aims to design an auxiliary system which can provide professional advising and help user to make standard postures with real-time vocal feedback.
The real-time vocal exercise posture detecting system uses three depth cameras to analyze exercise posture. The system provides two modes for different identities. The first mode is coach-mode, offering professional coach to record training films and mark mobile joints. The second mode is user-mode, offering real-time vocal suggestion when user is exercising.
The techniques involve capturing user’s joints coordinate, then calculating angles of each body part, and comparing to coach’s model. Simultaneously, the system records each body part to detect the completeness of user’s motion and offers vocal suggestion. Vocal suggestion would be analyzed by the decision tree designed by this thesis to ensure if the suggestion was correct. The precision of the system, from experiments, can reach 96.1% and the average time for processing a frame in each camera in same time is 25.3ms.

[1]台灣趨勢研究, 健身中心數據, http://www.twtrend.com/en/share_cont.php?id=62, July, 2018
[2]P. Uttarwar and D. Mishra, “Development of a Kinect-Based Physical Rehabilitation System”, 2015 Third International Conference on Image Information Processing(ICIIP), pp. 387-392, 2015
[3]維基百科, TOF原理, https://en.wikipedia.org/wiki/Time-of-flight_camera, July, 2018
[4]T客邦, 身體就是控制器, https://www.techbang.com/posts/2936-get-to-know-how-it-works-kinect, July, 2018
[5]D. Grest, J. Woetzel, and R. Koch, “Nonlinear Body Pose Estimation
from Depth Images”, In Proc. DAGM, 2005
[6]S. Knoop, S. Vacek, and R. Dillmann, “Sensor Fusion for 3D Human
Body Tracking with an Articulated 3D Body Model”, In Proc. ICRA,
2006
[7]Y. Zhu and K. Fujimura, “Constrained Optimization for Human Pose
Estimation from Depth Sequences”, In Proc. ACCV, 2007
[8]M. Siddiqui and G. Medioni, “Human Pose Estimation from a Single
View Point, Real-Time Range Sensor”, In CVCG at CVPR, 2010.
[9]C. Plagemann, V. Ganapathi, D. Koller, and S. Thrun, “Real-Time
Identification and Localization of Body Parts from Depth Images”, In Proc. ICRA, 2010
[10]CSDN, 人體背景分離圖片, https://blog.csdn.net/u014365862/article/details/46849309, July, 2018
[11]Shotton J, Sharp T, Kipman A, et al., “Real-Time Human Pose Recognition in Parts from Single Depth Images”, Communications of the ACM, 56(1): pp. 116-124, 2013
[12]S. Belongie, J. Malik, and J. Puzicha, “Shape Matching and Object Recognition Using Shape Contexts”, IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 2, no. 4, pp. 509-522, 2002
[13]D. M. Gavrila, “Pedestrian Detection from a Moving Vehicle”, in Proceedings of European Conference Computer Vision, Dublin, vol. 2, pp.37-49, 2000
[14]C. R. Wren, A. Azarbayejani, T. Darrell, and A. P. Pentland, “Pfinder: Real-Time Tracking of the Human Body”, IEEE Trans. Pattern Anal. Mach. Intell., vol. 19, no. 7, pp. 780–785, 1997
[15]H. Sidenbladh, “Detecting Human Ootion with Support Vector Machines,” in Proc. Int. Conf. Pattern Recognit. (ICPR), vol. 2, pp. 188– 191, 2004
[16]Ji-Ho Cho, In-Yeop Chang, Seung-man Kim, and Kwan H. Lee, “Depth Image Processing Technique for Representing Human Actors in 3DTV using Single Depth Camera”, 2007 3DTV Conference, 2007
[17]Hao Zhang, Christopher Reardon, and Lynne E. Parker, “Real-Time Multiple Human Perception with Color-Depth Cameras on a Mobile Robot”, IEEE Transactions on Cybernetics, vol. 43, no. 5, pp. 2168-2267, 2013
[18]Muhammad Hassan Khan, Kimiaki Shirahama, Muhammad Shahid Farid, and Marcin Grzegorzek, “Multiple Human Detection in Depth Images”, 2016 IEEE 18th International Workshop on Multimedia Signal Processing (MMSP), 2017
[19]Bingbing Ni, Yong Pei, Pierre Moulin, and Shuicheng Yan, “Multilevel Depth and Image Fusion for Human Activity Detection”, IEEE Transactions on Cybernetics, vol. 43, no. 5, pp. 1383-1394, 2013
[20]P. S. Madanayake, W. A. D. K. Wickramasinghe, H. P. Liyanarachchi, H. M. D. M. Herath, A. Karunasena, and T. D. Perera, “Fitness Mate: Intelligent Workout Assistant using Motion Detection”, 2016 IEEE International Conference on Information and Automation for Sustainability (ICIAfS)
[21]Ferda Ofli, Gregorij Kurillo, Štěpán Obdržálek , Ruzena Bajcsy, Holly Brugge Jimison, and Misha Pavel, “Design and Evaluation of an Interactive Exercise Coaching System for Older Adults: Lessons Learned”, IEEE Journal of Biomedical and Health Informatics, Vol.20, no.1, pp. 201-212, 2016
[22]Ching-Biau Tzeng, Tzuu-Shaang Wey, Yu-Tang Lee, and Jing-Ho Chen, “Study a Method to Automatically Examine the Sports Fitness By Image Recognition”, 2011 International Conference on Consumer Electronics, Communications and Networks (CECNet), pp. 3555-3559, 2011
[23]Liang Liu, Melody Yin, and Sanjay Mehrotra, “Preventing Venous thromboembolism in patients admitted by hospital using Microsoft Kinect V2”, 2015 IEEE Biomedical Circuits and Systems Conference (BioCAS), 2015
[24]Chung-Liang Lai, Ya-Ling Huang, Tzu-Kuan Liao, Chien-Ming Tseng, Yung-Fu Chen, and D. Erdenetsogt, “A Microsoft Kinect-based virtual rehabilitation system to train balance ability for stroke patients”, 2015 International Conference on Cyberworlds (CW), pp. 54-60, 2016
[25]鄭植諺(2012), 以深度影像為基礎之健身器材輔助系統設計, 國立交通大學電控工程研究所碩士論文
[26]G. L. Foresti, L. Marcenaro, and C. S. Regazzoni, “Automatic detection and indexing of video-event shots for surveillance applications”, IEEE Trans. Multimedia, vol.4, issue.4, pp.459-471,
2002.
[27]Microsoft, Kinect for Windows SDK v1.8, https://www.microsoft.com/en-us/download/details.aspx?id=40278, July, 2018
[28]CSDN, 骨骼數據圖片, https://blog.csdn.net/zouxy09/article/details/8161617, July, 2018
[29]Tsai, R.Y., “An Efficient and Accurate Camera Calibration Technique for 3D Machine Vision”. Proc. International Conference on Computer Vision and Pattern Recognition, pp. 364-374, 1986
[30]Adam Campbell, 2011, THE Men’s Health Big Book of Exercises, 1^sted, Taiwan：木馬文化
[31]CSDN, Kinect V1 圖片, http://www.cs.nccu.edu.tw/~whliao/hci2012/Kinect_Note_for_
Windows_SDK.pdf, July, 2018
[32]痞客邦, 肱二頭肌圖片, http://chris751201.pixnet.net/blog/post/6722003-%E8%82%B1%E4%BA%8C%E9%A0%AD%E8%82%8C%E8%82%8C%E8%85%B1%E7%82%8E%28bicep-brachii-tendinitis%29, July, 2018
[33]愛經驗, 肱三頭肌圖片, http://www.how01.com/post_eNaEKDW6W1edA.html, July, 2018
[34]維基百科, 三角肌圖片, https://zh.wikipedia.org/zh-tw/三角肌, July, 2018