過去近二十年，肥胖已被世界衛生組織列為一種慢性疾病，隨著兒童肥胖的比例逐漸上升，促進身體活動量成為肥胖管理重要的一環。基於預防勝於治療的道理，在嬰兒時期學習各種動作時就應同時培養健康的活動習慣，進而做好體重控制使身心皆能健康發展。準確的測量嬰兒活動量不僅有助於長期記錄嬰兒的活動狀況，還能從活動量的評估中了解到下列幾項影響身體健康的重要因素，如飲食與活動熱量消耗之間是否達到平衡、活動量是否足以達到健康的標準以及嬰兒個人的活動習慣是否優良。
針對成人已有許多的身體活動量測量方法，如問卷調查、代理人填表、加速度感測器、心跳記錄器、計步器與雙標水標示法等等。上述問卷填寫之作法不適用於長期且對象為嬰兒之活動量紀錄，而常用來測量日常活動量的所有接觸式設備皆不適合佩戴在嬰兒身上，因接觸式設備的佩戴會造成嬰兒活動時的諸多干擾。有鑑於此，本研究提出了視覺式嬰兒身體活動量監測系統，將接觸式設備改為PT IP camera，且提供的主要功能為監測嬰兒每個當下活動的代謝當量、估計其活動時間的熱量消耗與每個當下活動量的等級。本研究不僅著重於解決上述的問題達到活動量的監測功能，系統還引入了一般視覺式監視系統具備的基本功能。
本研究所提出的視覺式嬰兒活動量監測系統主要是透過tracking object initialization, infant tracking, PT IP camera control 以及physical activity measurement四個步驟來監測其活動量。首先，系統會利用codebook background subtraction演算法建立嬰兒的追蹤特徵，接著利用追蹤特徵在相鄰影格間搜尋嬰兒的所在位置。在嬰兒可能離開監控畫面時系統會控制鏡頭轉動，確保嬰兒長期存在監控畫面中。實驗時本研究架設一台PT IP camera於嬰兒的遊戲空間的至高處，拍攝其日常活動影像，活動影像包含嬰兒躺、坐、趴、爬行、學步以及身體各部位之運動，不同活動量等級之動作都在實驗結果中完整呈現。最後，系統從追蹤的影像上擷取多種嬰兒活動特徵，將特徵值經過整合與轉換得到嬰兒活動時每個當下的代謝當量值，此代謝當量值經由公式計算嬰兒活動的總消耗熱量，並評估當下的活動等級。
本研究改善了過去接觸式設備的缺點，使用視覺的方式監測嬰兒之活動量，同時具備了追蹤與控制鏡頭的功能讓系統能監控的範圍更廣。另外，針對不同鏡頭狀態所使用的活動量特徵也不相同，多元的活動量特徵讓系統在測量活動量時能更加準確。本研究設計了同嬰兒在同月齡之不同動作的活動量分析、不同嬰兒相同動作的活動量分析、同嬰兒不同月齡之相同動作的活動量分析、嬰兒與成人互動之活動量分析以及長時間活動量分析共五個實驗。實驗結果證實本系統可實現室內追蹤且控制鏡頭的能力且提供可靠且正確的身體活動量測量結果。在未來，長期記錄嬰兒身體活動狀況並建立常模，可以幫助照護者與醫師快速判斷嬰兒身體與動作發展的情形，以便得到正確的診斷與治療。

Over the past 20 years obesity has been considered to be a chronic disease. One useful way to avoid obesity is to manage and control individual physical activity, especially for the infants. Thus to developing reliable physical activity monitoring systems has been the issue of the study in recent year. To monitor physical activity accurately not only can record individual energy expenditure for obesity management, but also can figure out the significant factors affecting the personal health. The significant factors include the balance between energy intake and energy expenditure, the achievement of required physical activity to keep personal healthy, and the measurement of the quality of individual physical activities.
Currently, various measurement methods have been developed to measure the individual physical activity. One kind of these methods is to make a report, including self-report, proxy-report, and diary-report. However, the method cannot monitor the infant physical activity accurately since they cannot make a report themselves. Moreover, some sensors and techniques are also developed to help to record or calculate the physical activity, including heart rate monitor, pedometers, accelerometers and doubly labeled water. These sensors and techniques will make the infants very uncomfortable. Therefore, this study proposes a vision-based infant physical activity measurement system to estimate the infant metabolic equivalent, energy expenditure and activity levels. The proposed system not only measures the infant physical activity automatically, but also embeds some general functions of in-home monitoring systems.
The input videos of the proposed system is obtained from one PT IP camera which is set on the ceiling. And the proposed system consists of four major stages: tracking object initialization, infant tracking, PT IP camera control, and physical activity measurement. First, a codebook background subtraction algorithm is applied to extract the infant from the input frames and to construct a tracking feature model. Once the infant has been extracted, the system then tracks the infant by using the tracking feature model. Moreover, the system also predicts infant behaviors and controls the PT IP camera movement to avoid the infant crawling or walking out of the monitoring scope. Finally, the infant physical activity is evaluated automatically. In this study, the infant physical activity is divided into four levels, each of which may correspond to some infant behaviors such as lying, sitting, standing, kicking, limbs movement, torso movement, crawling, walking.
A series of experiments is designed to show the correctness and robustness of the proposed system. They are (1) the physical activity analysis of an infant doing different kinds of activities during a month; (2) the physical activity comparison between two infants doing the same kinds of activities; (3) the physical activity analysis of an infant doing the same kind of activities in different months; (4) the physical activity analysis of two infants which are interacting with adults; (5) the one-day physical activity analysis of two infants. The proposed system can help the construction of the norm of infant physical activity to help the doctors to diagnose the infant's health in the future.

[Who12] World health organization, “Global Health Risk: Mortality and Burden of Disease Attributable to Selected Major Risks,” pp. 10-11, 2012.
[Gre90] K. Greaves, M. Hoyt, and T. Baranowski, “Children’s Activity Rating Scale (CARS): Description and Calibration,” Research Quarterly for Exercise and Sport, vol. 61, no. 1, pp. 26-36, 1990.
[Lop11] P. Loprinzi and B. Cardinal, “Measuring Children’s Physical Activity and Sedentary Behaviors,” Journal of Exercise Science and Fitness, vol. 9, no. 1, pp. 15-23, 2011.
[Sir01] J. Sirard and R. Pate, “Physical Activity Assessment in Children and Adolescents,” Journal of Sports Medicine, vol. 31, no. 6, pp. 439-454, 2001.
[Sal85] J. Sallis, W. Haskell, P. Wood, S. Fortmann, T. Rogers, S. Blair and R. Paffenbarger, “Physical Activity Assessment Methodology in the Five-City Project,” American Journal Epidemiology, vol. 121, no. 1, pp. 91-106, 1985.
[Bou83] C. Bouchard, A. Tremblay, C. Leblanc, G. Lortie, R. Savard, and G.Theriault, “A Method to Assess Energy Expenditure in Children and Adults,” American Journal of Clinical Nutrition, vol. 37, pp. 461-467, 1983.
[Nei98] A. Neil, “Young People’s Physical Activity Patterns as Assessed by Heart Rate Monitoring,” Journal of Sports Sciences, vol. 16, pp. S9-S16, 1998.
[Bra07] D. Bravata, C. Smith-Spangler, V. Sundaram, A. Gienger, N. Lin, R. Lewis, C. Stave, I. Olkin, and J. Sirard, “Using Pedometers to Increase Physical Activity and Improve Health,” Journal of the American Medical Association, vol. 298, no. 19, pp. 2296-2304, 2007.
[Tre01] M. Tremblay, R. Shephard, T. McKenzie, and N. Gledhill, “Physical Activity Assessment Options within the Context of the Canadian Physical Activity, Fitness, and Lifestyle Appraisal,” Canadian Journal of Applied Physiology, vol. 26, no. 4, pp. 388-407, 2001.
[Yu12] M. Yu, A. Rhuma, S. Naqvi, L. Wang, and J. Chambers, “A Posture Recognition-Based Fall Detection System for Monitoring an Elderly Person in a Smart Home Environment,” Journal of Biomedical and Health Informatics, vol. 16, no. 6, pp. 1274-1286, 2012.
[Guo13] K. Guo, P. Ishwar, and J. Konrad, “Action Recognition from Video Using Feature Covariance Matrices,” IEEE Transactions on Image Processing, vol. 22, no. 6, pp. 2479-2494, 2013.
[Ali10] S. Ali and M. Shah, “Human Action Recognition in Videos Using Kinematic Features and Multiple Instance Learning,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 32, no. 2, pp. 288-303, 2010.
[Der13] K. Derpanis, M. Sizintsev, K. Cannons, and R. Wildes, “Action Spotting and Recognition Based on a Spatiotemporal Orientation Analysis,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 35, no. 3, pp. 527-540, 2013.
[Bru12] D. Brulin, Y. Benezeth, and E. Courtial, “Posture Recognition Based on Fuzzy Logic for Home Monitoring of the Elderly,” IEEE Transactions on Information Technology in Biomedicine, vol. 16, no. 5, pp. 974-981, 2012.
[Mir13] B. Mirmahboub, S. Samavi, N. Karimi, and S. Shirani, “Automatic Monocular System for Human Fall Detection Based on Variations in Silhouette Area,” IEEE Transactions on Biomedical Engineering, vol. 60, no. 2, pp. 427-436, 2013.
[Sta12] A. Stahl, C. Schellewald, Ø. Stavdahl, O. Aamo, L. Adde, and H. Kirkerød, “An Optical Flow-Based Method to Predict Infantile Cerebral Palsy,” IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol. 20, no. 4, pp. 605-614, 2012.
[Osa09] Y. Osawa, K. Shima, N. Bu, T. Tsuji, T. Tsuji, I. Ishii, H. Matsuda, K. Orito, T. Ikeda, and S. Noda, “A Motion-Based System to Evaluate Infant Movements Using Real-Time Video Analysis,” Proceedings of the International Conference on Biomedical Engineering, Venice, Italy , pp. 2043-2047, 2009.
[Bou13] N. Boulgouris and X. Huang, “Gait Recognition Using HMMs and Dual Discriminative Observations for Sub-Dynamics Analysis,” IEEE Transactions on Image Processing, vol. 22, no. 9, pp. 3636-3647, 2013.
[Kim05] K. Kim, T. Chalidabhongse, D. Harwood, and L. Davis, “Real-time Foreground–Background Segmentation Using Codebook Model,” Journal of Real-Time Imaging, vol. 11, no. 3, pp. 172-185, 2005.
[Bra08] G. Bradski and A. Kaehler, “Learning OpenCV: Computer Vision with the OpenCV Library,” O’Reilly, pp. 267-315, 2008.
[Tea80] M.R. Teague, “Image Analysis via the General Theory of Moments,” Journal of Optical Society of America, vol. 70, no. 8, pp. 920-930, 1980.
[Erd01] C. E. Erdem, A. M. Tekalp and B. Sankur, “Metrics for Performance Evaluation of Video Object Segmentation and Tracking without Ground-Truth,” Proceedings of International Conference on Image Processing, Thessaloniki, Greece, vol. 2, pp. 69-72, 2001.
[Pre92] W. H. Press, S. A. Teukolsky, W. T. Vetterling, and B. P. Flanney, “NumericaL Recipes in C,” Cambridge Univeristy Press, pp. 620-623, 1992.
[Bob96] A. Bobick and J. Davis, “Real-Time Recognition of Activity Using Temporal Templates,” Proceedings of IEEE Workshop on Applications of Computer Vision, pp. 39-42, 1996.
[Dav99] J. Davis and A. Bobick, “Real-Time Motion Template Gradients Using Intel CVLib,” Proceedings of IEEE International Conference on Computer Vision Workshop on Framerate Vision, pp. 1-20, 1999.
[Bra00] G. Bradski and J. Davis, “Motion Segmentation and Pose Recognition with Motion History Gradients,” Proceedings of IEEE Workshop on Applications of Computer Vision, Palm Springs, CA, pp. 238-244, 2000.
[Xav02] F. Xavier, “The Obesity Epidemic: Pathophysiology and Consequences of Obesity,” Journal of Obesity Research, vol. 10, no. 2, pp. 97S-103S, 2002.
[Fie01] A. E. Field, E. H. Coakley, J. L. Spadano, N. Laird, W. H. Dietz, E. Rimm and G. A. Colditz, “Impact of Overweight on the Risk of the Risk of Developing Common Chronic Diseases During a 10-Year Period,” Journal of Archives of Internal Medicine, vol. 161, no. 13, pp. 1581-1586, 2001.
[Wal14] W. P. Walker and D. K. Bhatia, “Automated Ingestion Detection for a Health Monitoring System,” Journal of Biomedical and Health Informatics, vol. 18, no. 2, pp. 682-692, 2014.
[Edg13] A. Edgcomb and F. Vahid, “Estimating Daily Energy Expenditure from Video for Assistive Monitoring,” Proceedings of IEEE International Conference on Healthcare Informatics, Sheraton Philadelphia University, USA, pp. 184-191, 2013.
[張07] 張美惠，〈肥胖防治要從小紮根〉。出自兒童健康推展委員會，2007。
[祝04] 祝年豐，〈兒童肥胖症〉。出自肥胖核心課程系列-小兒肥胖症，2004。
[卓02] 卓玉蓮，〈應用跨理論模式促進國中學生身體活動：認知-行為教育介入〉。出自國立臺灣師範大學碩士論文，2002。
[邵09] 邵柏潤，〈以肢體動作分析為基礎之新生兒意外監控系統〉。出自國立臺灣師範大學碩士論文，2009。
[蕭10] 蕭宛甄，〈以前景物動態機率模型為基礎之嬰兒危險程度評估系統〉。出自國立臺灣師範大學碩士論文，2010。
[江09] 江瑞坤、林名男、黃郁文，〈代謝當量簡介〉。基層醫學，第二十四卷，第三期，第112-115頁，2009。
[1] International Obesity Task Force, “Global Overweight in children,” Available at: http://www.iaso.org/resources/obesity-data-portal/resources/
tables/, Accessed 2013.
[2] International Association for the Study of Obesity, “World map of obesity,” Available at: http://www.worldobesity.org/, Accessed 2011.
[3] 行政院衛生署國民健康局。〈兒童動作發展〉。出自兒童健康手冊。2011。
[4] 無憂寶寶。0-3歲寶寶增強體質的身體運動。網址：http://www.baby910.c
om/。上網日期：2013-08-05。
[5] JW-005T三功能健康計步器。網址：http://pedometer.tw.rakuten-static.com /p10191295947-1.jpg。上網日期：2013-08-05。
[6] TIMEX心跳卡路里計測運動錶。網址：http://rp1.monday.vip.tw1.yahoo.n
et/res/gdsale/st_pic/4254/st-4254447-s400.jpg。上網日期：2013-08-05。
[7] RT3 Tri-axial。網址：http://www.pt.ntu.edu.tw/labs/lxjst/equip_1.jpg。上網日期：2013-08-05。
[8] 行政院衛生署。〈國人膳食營養素參考攝取量修訂第七版〉。出自國人膳食營養素參考攝取量修訂第七版。2011。