許永真臺灣大學：資訊工程學研究所潘昭瑋Pan, Chao-WeiChao-WeiPan2007-11-262018-07-052007-11-262018-07-052007http://ntur.lib.ntu.edu.tw//handle/246246/54127乳癌患者在手術完畢後，必須要進行一系列的復健動作，然而病患在家中自行進行復健動作的過程中，常常因為疼痛，或種種因素而使得整個復健動作流程不完整或中斷。本研究透過動作辨識來紀錄病患每天在家中的復健情形將會對醫生在門診時能更快速的了解病人復健狀況，給予病人更有效的復健療程。 本篇論文主要採用人體在進行復健動作過程中的加速度作為訊號來源，利用機器學習(Machine Learning)的方法，針對八種基本的乳癌復健動作進行辨識。本文探討了加速度訊號的重要特徵值，並以隱藏式馬可夫模型(Hidden Markov Model)為理論基礎進行一系列的實驗來驗證其辨識的準確率。實驗結果指出，在使用平均值、能量、頻譜亂度、相關係數當特徵值時所得到的模型可達到98%的準確性。Rehabilitation exercises after breast cancer surgery can help prevent post-operation complications. This thesis adopts activity recognition technique to identify and record patients' rehabilitative exercises. This information helps doctors monitor the patients' conditions in follow-up visits. This thesis presents a activity recognition system based on continuous hidden Markov models. Accelerometers are used to capture the upper body movements when patients do rehabilitation. Four different representative features, mean, energy, entropy, and correlation, are extracted from signals. The recognition rate of exercises is about 98%. The performance of the recognizer is also evaluated in both user dependent and user independent cases.Acknowledgments ii Abstract iii List of Figures viii Chapter 1 Introduction 1 1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 Problem Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Chapter 2 RelatedWork 5 2.1 Detection of Ambulatory Mode . . . . . . . . . . . . . . . . . . . . . 6 2.2 ADL Recognition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 2.3 Gesture Recognition . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Chapter 3 Experiment Design 9 3.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 3.2 Accelerometers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 3.2.1 Hardware . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 3.2.2 Placement . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 3.3 Target Rehabilitation Exercises . . . . . . . . . . . . . . . . . . . . . 14 3.4 Proposed Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 3.5 Feature Extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 3.6 Recognition Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . 21 Chapter 4 Experiment and Evaluation 25 4.1 Data Collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 4.2 Experiment Results . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 Chapter 5 Conclusion 33 Bibliography 35 Appendix A The Breast Cancer Rehabilitation Exercise 371601711 bytesapplication/pdfen-US加速度動作辨識復健AccelerationActivity RecognitionRehabilitation[SDGs]SDG3thesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/54127/1/ntu-96-R94922141-1.pdf