本論文主旨在以配戴加速度感測模組擷取與分析人體活動之加速度訊號變化來實現人體動作辨識，利用配戴於腳踝之三軸加速度感測模組紀錄騎腳踏車、走路、上樓和下樓這四種活動加速度訊號，分別經由波形分析演算法和小波分析演算法進行動作辨識。
在波形分析演算法裡，騎腳踏車時加速度感測模組所產生的加速度訊具有以下的特性：1) 垂直於地面軸向(X軸)的加速度訊號近似於正弦波，其波型振幅上下震盪的程度十分相近，2) 在頻譜圖上有一明顯的主頻，此主頻的週期反映在加速度訊號上的波峰與波谷的距離上。因此藉由計算出原始垂直軸加速度訊號之平均振幅作為估計主頻振幅大小並找出加速度訊號的波峰與波峰以及波谷與波谷的距離推算出主頻頻段，當頻譜圖的主頻落在主頻頻段裡且主頻的振幅大於估計主頻振幅大小則加速度訊號會被辨識為腳踏車，其平均辨識率高達97 %。
在小波分析演算法裡，藉由分析三軸加速度訊號所計算出的signal vector magnitude (SVM)來辨識走路、上樓與下樓。SVM訊號在接著經過小波轉換分解出一到四層的細節訊號(detail signal) ，通過特徵擷取計算出33種特徵值當作類神經網路的輸入，在經過類神經網路的訓練和測試之後其辨識率可達到91 %。由實驗證實本論文所提出之動作辨識的有效性。

This thesis develops a physical activity classification algorithm using acceleration signals generated from an accelerometer module. The subjects wear a triaxial accelerometer module on the ankle to measure and collect the acceleration signals of riding a bicycle, level walking, walking upstairs, and walking downstairs. The proposed classification algorithm based on a waveform and wavelet analysis is used to classify four activities.
In the waveform analysis algorithm, the acceleration signals of riding a bicycle have the following characteristics: 1) The vertical acceleration signals are like a sine wave and the amplitudes of each peak and valley are almost equal; 2) the spectrums of Fourier transform have one obvious frequency (the main frequency) and the period of main frequency reflect to the distance between peak and peak or valley to valley. From the first characteristic, we used the amplitude of vertical acceleration signals to estimate the main frequency magnitude. From the second characteristic, we detect the peaks and valleys of the vertical acceleration signals and then compute the averaged inter-peak and inter-valley distance to estimate the frequency band of main frequency. The acceleration signals are classified as bicycling if the main frequency is located in frequency band, and the main frequency magnitude is greater than estimated frequency magnitude. The classification rate of bicycling can achieve 97%.
In the wavelet analysis algorithm, we analyze the signal vector magnitude (SVM) signals generated from triaxial acceleration signals to discriminate level walking, walking downstairs, and walking upstairs. The SVM signals are decomposed into four-level detail signals by wavelet transformation. 33 features as inputs of the classifier are computed by the SVM and detail signals. A feedforward neural network is adopted as the classifier. The classification rate of three activities can achieve 91%. The experimental results can validate the effectiveness of the proposed physical activity algorithm.

[1] I. Haritaoglu, D. Harwood, and L. S. Davis, “Real-time surveillance of people and their activities,” IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 22, no. 8, pp. 809-830, 2000.
[2] M. Engin, S. Demirağ, E. Z. Engin, G. Çelebi, F. Ersan, E. Asena, and Z. Çolakoğlu, “The classification of human tremor signals using artificial neural network,” Expert Systems with Applications, vol. 33, pp. 754-761, 2007.
[3] M. Zimmermann and H. Bunke, “Hidden markov model length optimization for handwriting recognition systems,” Eighth International Workshop on Frontiers in Handwriting Recognition, pp. 369-374, 2002.
[4] J. J. Lee, J. Kim, and J. H. Kim, “Data-driven design of HMM topology for online handwriting recognition” International Journal of Pattern Recognition and Artificial Intelligence, vol. 15, pp.107-121, 2001.
[5] K. T. Song and Y. Q. Wang, “Remote activity monitoring of the elderly using a two-axis accelerometer,” Proceedings of 2005 CACS Automatic Control Conference, vol. 22, no. 8, 2000, pp.18-23.
[6] M. J. Mathie, A. C. F. Coster, N. H. Lovell, and B. G. Celler, “Accelerometry: providing an integrated, practical method for long-term, ambulatory monitoring of human movement,” Physiological Measurement, vol. 25, pp. R1-R20, 2004.
[7] A. Godfrey, R. Conwaya, D. Meagherd, and G. ÓLaighin, “Direct measurement of human movement by accelerometry,” Medical Engineering & Physics, vol. 30, pp. 1364-1386, 2008.
[8] M. Sekine, T. Tamura, M. Ogawa, T. Togawa, and Y. Fukui, ”Classification of acceleration waveform in a continuous walking record,” in Proceeding of IEEE 20th Annual International Conference on Engineering in Medicine and Biology Society, vol. 20, no. 3, 1998, pp. 1523-1526.
[9] M. Sekine, T. Tamura, T. Togawa, and Y. Fukui, “Classification of waist-acceleration signals in a continuous walking record,” Medical Engineering & Physics, vol. 22, pp. 285-291, 2000.
[10] M. Sekine, T. Tamura, T. Fujimoto, and Y. Fukui, “Classification of walking pattern using acceleration waveform in elderly people,” in Proceedings of 22nd Annual EMBS International Conference, vol. 2, 2000, pp. 1356-1359.
[11] M. Sekine, T. Tamura, M. Akay, T. Fujimoto, T. Togawa, and Y. Fukui, “Discrimination of walking patterns using wavelet-based fractal analysis,” IEEE Trans. on Neural Systems and Rehabilitation Engineering, vol. 10, pp. 188-196, 2002.
[12] T. Tamura, M. Sekine, M. Ogawa, T. Togawa, and Y. Fukui, “Classification of acceleration waveforms during walking by wavelet transform,” Methods of Information in Medicine, vol. 36, pp. 356-359, 1997.
[13] Ning Wang, N. H. Lovell, E. Ambikairajah, and B. G. Celler, “Accelerometry based classification of walking patterns using time-frequency analysis,” in Proceeding of IEEE 29th Annual International Conference on Engineering in Medicine and Biology Society, 2007, pp. 4899-4902.
[14] M. N. Nyan, F. E. H. Tay, K. H. W. Seah, and Y. Y. Sitoh, “Classification of gait patterns in the time–frequency domain,” Journal of Biomechanics, vol. 39, pp. 2647-2656, 2006.
[15] S. J. Preece, J. Y. Goulermas, L. P. J. Kenney, and D. Howard, “A comparison of feature extraction methods for the classification of dynamic activities from accelerometer data,” IEEE Trans. Biomedical Engineering, vol. 56, pp. 871-879, 2009.
[16] J. Pärkkä, L. Cluitmans, and M. Ermes, “Personalization algorithm for real-time activity recognition using PDA, wireless motion bands, and binary decision tree,” IEEE Trans. Information Technology in Biomedicine, vol. 14, pp.1211-1215, 2010.
[17] S. J. Preece, J. Y. Goulermas, L. P. J. Kenney, D. Howard, K. Meijer, and R. Crompton, “Activity identification using body-mounted sensors—a review of classification techniques,” Physiological Measurement, vol. 30, no. 4, pp. R1-R33, 2009.
[18] C. V. C. Bouten, K. T. M. Koekkoek, M. Verduin, R. Kodde, and J. D. Janssen, “A triaxial accelerometer and portable data processing unit for the assessment of daily physical activity,” IEEE Trans. Biomedical Engineering, vol. 44, pp. 136-147, 1997.
[19] E. K. Antonsson and R. W. Mann, “The frequency content of gait,” Journal of Biomechanics, vol. 18, pp. 39-47, 1985.
[20] B. E. Ainsworth, W. L. Haskell, M. C. Whitt, M. L. Irwin, A. M. Swartz, S. J. Strath, W. L. O'Brien, D. R. Bassett, K. H. Schmitz, P. O. Emplaincourt, D. R. Jacobs, and A. S. Leon, “Compendium of physical activities: an update of activity codes and MET intensities,” Medicine and Science in Sports and Exercise, vol. 32, pp. S498-504, 2000.
[21] R. M. Rangayyan, Biomedical Signal Analysis: A Case-Study Approach. Wiley Inter-Science, 2011.
[22] T. Chau, “A review of analytical techniques for gait data Part2 neural network and wavelet methods,” Gait & Posture, vol. 13, pp. 102-120, 2001.
[23] D. M. Karantonis, M. R. Narayanan, M. Mathie, N. H. Lovrll, and B. G. Celler, “Implementation of a real-time human movement classifier using a triaxial accelerometer for ambulatory monitoring,” IEEE Trans. Information Technology in Biomedicine, vol. 10, no. 1, pp. 156-167, 2006.
[24] 林傑斌、陳湘、劉明德，SPSS 11統計分析實務設計寶典，博碩，民國91年4月。
[25] 張紹勳、林秀娟，SPSS FOR WINDOWS統計分析-初等統計與高等統計 (下冊)，松崗文魁，民國89年5月。
[26] 陳順宇，多變量分析(四版)，華泰文化，民國94年6月。
[27] Y. P. Chen, J. Y. Yang, S. N. Liou, G. Y. Lee, and J. S. Wang, “Online classifier construction algorithm for human activity detection using a tri-axial accelerometer,” Applied Mathematics and Computation, vol. 25, pp. 849-860, 2008.
[28] 周文賢，統計學，智勝，民國94年1月。
[29] J. B. Bussmann, J. H. Tulen, E. C. van Herel, and H. J. Stam, “Quantification of physical activities by means of ambulatory accelerometry: a validation study,” Psychophysiology, pp. 488-496, 1998.
[30] J. Pärkkä, M. Ermes, K. Antila, M. van Gils, A. Mänttäri, and H. Nieminen, “Estimating Intensity of Physical Activity: A Comparison of Wearable Accelerometer and Gyro Sensors and 3 Sensor Locations,” in Proceedings of IEEE 29th Annual International Conference on Engineering in Medicine and Biology Society, 2007, pp. 1511-1514.
[31] J. G. Barton and A. Lees, “Development of a connectionist expert system to identify foot problems based on under-foot pressure patterns,” Clinical Biomechanics, vol. 10, no. 7, pp. 385-391, 1995.