即時行為模式辨識與評估系統之研製

Design and Implementation of Real-Time System for Behavior Model Evaluation and Classification

劉憲璁(Hsien-Chung Liu)

指導教授 ： 李仁貴

國立臺北科技大學/電資學院/電腦與通訊研究所/碩士(2010年)

https://doi.org/10.6841/NTUT.2010.00043

摘要

人口老化所帶來的嚴重問題，以及在疾病型態的慢性化下，使得需要長期健康照護的人口隨之增加。本文係針對高齡者的身體活動能力狀況之照護需求為出發，藉由長期的監測及記錄身體活動資料，作為後續的評估分析，達到防患與即時治療，降低身體功能退化及罹病率，維持其自主和獨立性，並享有高品質的生活。
本論文透過娛樂及照護需求之整合應用，結合當下流行的任天堂遊戲機－Wii Remote開發出一套即時性的日常生活行為模式監控、辨識與評估系統。藉由在使用者身上配戴Wii Remote裝置，利用其三軸加速度計所產生的資訊，經由系統設計的行為辨識演算法，來推測使用者的身體活動情況，並即時的監測與記錄活動資料。系統可透過使用者互動介面來觀測在日常生活中的動作轉換、身體姿態以及行走等身體活動情形，且加強在意外跌倒事件上的偵測設計，提升其預警與辨識功能。從實驗的結果驗證，系統在身體姿態(坐、臥、躺與站)的判別和行走情形的偵測，皆具有良好的辨識準確度，平均辨識準確率為97%。本系統實地應用在生活環境中進行長時間的監測，並評估使用者的身體活動情形，由結果顯示可完整且精確地紀錄其日常行為活動資料。
最後，在未來的應用發展方面，系統可加強在行為辨識演算法的自我學習與適應能力上，以及透過3D使用者互動介面來呈現日常生活的動作狀態，並可再拓展多元的照護需求，以達到無所不在之照護服務為願景。

關鍵字

身體活動 ； 日常生活活動 ； 三軸加速度計 ； Wii Remote ； 即時性

並列摘要

The aging of the population is the serious problems and the trend of chronic diseases, making the long-term health care needs of the population increased. This article is for the elderly physical activity (PA) status of the care needs as the starting, through the long-term physical activity monitoring and recording data as a follow-up assessment, and reach the precaution and timely treatment of early-warning, reduce the degradation of physical function and morbidity to maintain their autonomy and independent, and enjoy the high quality of life (QOL).
This paper through the integration of entertainment and care needs to combination of current popular Nintendo－Wii Remote, and to develop a real-time monitoring of behavior identification and evaluation system. That used the Wii Remote attached to the user's body and through its tri-axial accelerometer information and using the behavior recognition algorithms to detect the user's physical activities and real-time monitoring and recording information on their activities. This system could through the user interaction interface observed to the activities of daily living (ADLs) such as the movement transitions, body postures and walking and other physical activities. In addition, to strengthen the detection of accidental fall incident designed to improve its warning and recognition. The results for evaluating the performance of the system show the high identification accuracy for the body postures (sitting, lying and standing) and walking, and the average recognition accuracy rate of 97%. The system used the long-term test of life environment to monitor and evaluation activities of the user's physical situation, the results can be complete and accurate records of information on the activities of daily living.
In the future of the application development, this system could enhance the self-learning and adaptation ability of the behavior recognition algorithms, and through the 3D user interface (3DUI) to show of their activities of daily living (ADLs), and may further expand the care needs of diverse to achieve the prospects of the ubiquitous healthcare services.

並列關鍵字

Physical activity ； Activities of daily living ； Tri-axial accelerometer ； Wii Remote ； Real-time

參考文獻

[45] WHO, International Classification of Functioning, Disability and Health (ICF) [Online]. Available: http://www.who.int/classifications/icf/en/

[1] World health organization of the United Nations, Geneva, Switzerland, [Online]. Available: http://www.who.int (last date visited: Jul. 23, 2008).

[5] J. W. Rowe and R. L. Kahn, Successful Aging, The gerontologist, 1997.

[6] M. Galloway and P. Jokl, "Aging successfully: The importance of physical activity in maintaining health and function," Journal of the American Academy of Orthopaedic Surgeons, vol. 8, no. 1, 2000, pp. 37-44.

[8] J. Fries, "Aging natural death and the compression of morbidity," New English Journal of Medicine, vol. 303, no.3, 1980, pp. 130-135.