臺灣博碩士論文加值系統

復健是需要長期的投入時間來練習，也隨之帶來醫療人員短缺的問題，而若可以把維持性的復健時間轉移到家裡，這樣一方面可紓解復健者家人的壓力，一方面也可達到舒緩醫療人員短缺之問題，因此本研究實現出一套居家上半身復健系統。上半身的肢體辨識，最大的關鍵就是如何穩定的去建立出上半身的骨架點，而本研究提出一個穩定且快速的上半身骨架點建立演算法，透過的上半身人體骨架化以及人體膚色的檢測，相輔相成的建立出上半身骨架點，而本研究對於所定義好的復健動作辨識率可以達到98%，代表著本研究判斷的準確性，其中定義的復健動作皆有其復健肌肉部位說明，而且一張畫面只需要8ms就可以完成上半身人體骨架點的建立，相當於可提供125FPS的速度，這樣速度可以讓本研究在以後的擴充上可以有著更大的空間。本研究也另外嘗試了採用單一RGB攝影機去取代Kinect的深度，一樣可以得到有效的結果，雖然使用環境較為受限，但這對於未來本研究移植到嵌入式開發平台上墊定好基礎。

Rehabilitation takes a lot of time to practice rehabilitation actions, and thus causing a shortage of medical and rehabilitation staffs. As a result, home rehabilitation can not only reduce the loadings of family members of rehabilitation pedestrian, but also relieve shortages of medical and rehabilitation staffs, this thesis proposes an action identification of upper body rehabilitation system. For the proposed system, which is the most important point of building upper body skeleton, this thesis presents an algorithm to feasibly and rapidly build upper body skeleton points. Through the upper body of the human skeleton and human skin color information, an upper body skeleton points can be effectively established by the proposed system. As a result, the proposed system can achieve a high recognition rate of 98% for the defined rehabilitation actions for different muscle parts. Moreover, the computational speed of the proposed system can reach 125 FPS, i.e. the processing time per frame is 8ms, the computational efficiency can provide efficient extensibility in the future development for dealing with the complex ambient environments and the implementation on embedded and pervasive systems.

摘　要 I
ABSTRACT II
誌 謝 IV
目 錄 V
表目錄 VIII
圖目錄 IX
第一章 緒論 1
1.1 研究動機 1
1.2 研究目的 1
1.3 章節提要 2
1.4 論文貢獻 3
第二章 文獻回顧 5
2.1 復健背景 5
2.2 膚色 6
2.2.1 RGB色彩空間 6
2.2.2 HSV色彩空間 6
2.2.3 YCbCr 色彩空間 7
2.3 骨架 7
2.3.1 MS SDK 8
2.3.2 OpenNI + NITE 8
2.3.3 輪廓 9
2.4動作判定 10
第三章 研究方法 12
3.1 實驗環境 12
3.2 系統流程 15
3.3膚色 17
3.3.1 色彩空間的選擇 17
3.3.2 橢圓膚色模型的應用 21
3.3.3 膚色檢測流程 22
3.4骨架 23
3.4.1人體骨架化 24
3.4.2 骨架-頭部 31
3.4.3 骨架-肩膀 32
3.4.4 骨架-手 34
3.4.5 骨架-手肘 34
3.4.6 骨架-旋轉骨架點修正 35
3.5 時間彎折演算法(DYNAMIC TIME WARPING) 37
3.6 採用RGB攝影機實現方法 39
第四章 實驗結果與分析 42
4.1軟硬體環境介紹 42
4.2 系統執行速度 43
4.3 膚色偵測的結果 45
4.4 骨架建立效果 50
4.5時間彎折演算法(DYNAMIC TIME WARPING)判斷結果 53
4.6 採用RGB攝影機的效果 56
4.7文獻比較 58
第五章 結論 66
參考文獻 68

[1]Helen J. Hislop, Daniels and Worthingham’s MUSCLE TESTING Techniques of Manual Examination 8th Edition, Elsevier Science Health Science Division, 2007, pp. 96-115
[2]Wikipedia維基百科, "HSL與HSV色彩空間" https://zh.wikipedia.org/wiki/HSL%E5%92%8CHSV%E8%89%B2%E5%BD%A9%E7%A9%BA%E9%97%B4
[3]cnKinect.com, "Kinect for Windows 骨架追踪"
http://www.cnkinect.com/
[4]Heresy’s Space, "透過 OpenNI / NITE 分析人體骨架（上）" http://kheresy.wordpress.com/2011/01/28/detecte_skeleton_via_openni_part1/
[5]Chi-Min Oh, "Pictorial Structures-based Upper Body Tracking and Gesture Recognition," 2011 17th Korea-Japan Joint Workshop on Frontiers of Computer Vision (FCV), Ulsan, 2011, pp. 1-6
[6]Georg Hackenberg, "Lightweight Palm and Finger Tracking for Real-Time 3D Gesture Control," 2011 IEEE Virtual Reality Conference (VR), Singapore, 2011, pp. 19-26
[7]T. Gonzalez-Sanchez, "Real-time body gesture recognition using depth camera," ECTRONICS LETTERS, vol.47, no.12, 2011, pp. 697-698
[8]Heresy’s Space, "體感操作與 OpenNI 介紹"
http://kheresy.wordpress.com/tag/openni/
[9]C. Patlolla, "Real-time hand-pair gesture recognition using a stereo webcam," 2012 IEEE International Conference on Emerging Signal Processing Applications (ESPA), NV Las Vegas, 2012, pp. 135-138
[10]Chuan-xu Wang, "Face Detection Based on Skin Gaussian Model and KL Transform," 2011 International Symposium on Computer Science and Society (ISCCS), Kota Kinabalu, 2011, pp. 100-103
[11]Sung-il Kang , "Using Depth and Skin Color for Hand Gesture Classification," 2011 IEEE International Conference on Consumer Electronics (ICCE), NV Las Vegas, 2011, pp. 155-156
[12]Google 圖片, "黑人"
http://www.google.com.tw/search?hl=zh-TW&site=imghp&tbm=isch&source=hp&biw=1536&bih=728&q=%E9%BB%91%E4%BA%BA&oq=%E9%BB%91%E4%BA%BA&gs_l=img.3...4125.5344.0.5677.3.3.0.0.0.0.215.292.2j0j1.3.0...0.0.0..1ac.1j4.17.img.k2j_Ubi0Ds4
[13]Spottr Labs, "A super-simple skin detector in OpenCV"
http://spottrlabs.blogspot.tw/2012/01/super-simple-skin-detector-in-opencv.html