隨著醫療水準的大幅提升及自然出生率的下降，台灣社會已呈現高齡化的狀態，雖然人的壽命延長，但罹患各種慢性病的機會也伴隨著年齡而增加，照護中心、安養機構等單位隨之成立，如何藉著日新月異的科技發展來改善老年人的日常生活甚至提供健康照護上的協助，已經是目前相當重要的議題。
為了監看老年人的日常生活與保障他們的生命安全，裝設視訊監控系統是大多數照護中心採用的方式。現今社會非常注重隱私權，以醫療照護系統為例，若在一些無法使用全彩攝影機的單位或情境下，必須考慮患者與人員的隱私權益問題。彩色影像的系統雖然能達到高辨識率，卻無法保護個人隱私。目前本實驗室已發展出一套使用熱感影像的睡眠評估系統，不僅能夠降低隱私權的問題，也能夠達到基本上下床的偵測以進行睡眠評估。但在熱感影像睡眠評估系統中，僅能夠取得人體的溫度分佈與熱源輪廓，尚未能夠對其影像進行身分辨識，若有非原房間住民進入時，會導致辨識與偵測上的誤差，抑或是照護中心發生不同住民誤入他人房間時，可能會造成他人的權益受損。
緣此，本研究主要目標為發展出一套基於熱感影像的照護中心住民身分辨識演算法，利用擷取運動熱感移動影像的人體移動特徵，結合深度學習的技術與架構，最終達到住民身分辨識的目的。經實驗結果證明，本論文提出的方法，在熱感影像中作物件偵測的準確率可以達到97%，以熱感影像辨識住民是否為所屬房間的辨識率可以達到85%的效果，透過實驗的結果能夠驗證本論文提出方法之有效性。筆者期許這樣的身分辨識架構能夠有效的幫助到無法使用全彩監控系統的單位或情境，提供他們能夠更準確的掌握住民或患者的情況。

Nowadays, with the substantial increase in medical standards and the decline in mortality rate, human life expectancy is extended. In Taiwan, low birth rate and mortality rate are the main reasons that our country has become an aging society. Although human life is prolonged, the chances of suffering from chronic diseases are accompanied by age. With this situation, medical institutions and care centers have been established year after year. Therefore, how to improve the daily life of the elderly and provide them with health care assistance through state-of-the-art technology has already been a social issue of high concern.
In order to monitor the condition of the elderly and to ensure their safety and protect their lives, the installation of video surveillance systems is the way that most care centers used. In recent years, our society is concerned more about the privacy. For example, when using medical care system, medical caregivers must take the privacy of patients and medical staff into consideration. Video surveillance system with color image can achieve high recognition rate to know patients’ conditions, but it cannot protect individual privacy. At present, our laboratory has developed a thermal image based sleep summarization system. It not only reduces the privacy issues, but also achieves the basic sleep assessment. However, in this system, we can only acquire the body’s temperature distribution and heat source contour, and still fail to recognize the identity of the person in the image. If non-original room inhabitants walk into the room, such as caregivers or other inhabitants, it will lead to the error of identification and detection of this system and may cause damage to right of others.
Hence, the main purpose of this research is to develop an identification algorithm based on thermal imaging. It can achieve the purpose of recognizing identification by capturing the human movement feature in the thermal moving images and combining the techniques and structure of machine learning. The experiment shows that the proposed object detection algorithm on thermal image has satisfied to 97%, and the proposed algorithm has satisfied detecting rate up to 85% for identifying whether the person walking into the room is the original inhabitant or not. Through the experimental results, we can verify the effectiveness of the proposed method. The identity of the identification algorithm can not only effectively help those institutions or situations, which cannot use full-color monitoring system, but also provide caregivers with more efficient grasp of the situations of the inhabitants or patients.

[1] 林莉莉，《臺灣高齡化社會的可行人口發展規劃:需求層次觀點》，2013。
http://dx.doi.org/10.6581/lsgc.2013.66%2811%29.04
[2] 內政部統計處，《106年第10週內政統計通報》，2017。
http://sowf.moi.gov.tw/stat/week/week10610.pdf
[3] G. Ginesu, D. D. Giusto, V. Margner, and P. Meinlschmidt, "Detection of foreign bodies in food by thermal image processing," IEEE Transactions on Industrial Electronics, vol. 51, no. 2, pp. 480-490, 2004.
[4] I. Pavlidis, J. Levine, and P. Baukol, "Thermal imaging for anxiety detection," in Computer vision beyond the visible spectrum: Methods and applications, 2000. Proceedings. IEEE workshop on, 2000, pp. 104-109: IEEE.
[5] D. Zhou, M. Dillon, and E. Kwon, "Tracking-based deer vehicle collision detection using thermal imaging," in Robotics and Biomimetics (ROBIO), 2009 IEEE International Conference on, 2009, pp. 688-693: IEEE.
[6] B. C. Arrue, A. Ollero, and J. M. De Dios, "An intelligent system for false alarm reduction in infrared forest-fire detection," IEEE Intelligent Systems and Their Applications, vol. 15, no. 3, pp. 64-73, 2000.
[7] C.-L. Chiang, S.-H. Liu, Y.-C. Lin, and C.-M. Shu, "Thermal release hazard for the decomposition of cumene hydroperoxide in the presence of incompatibles using differential scanning calorimetry, thermal activity monitor III, and thermal imaging camera," Journal of Thermal Analysis and Calorimetry, vol. 127, no. 1, pp. 1061-1069, 2017.
[8] S. K. Biswas and P. Milanfar, "Linear support tensor machine with LSK channels: pedestrian detection in thermal infrared images," IEEE transactions on image processing, 2017.
[9] V. John, S. Mita, Z. Liu, and B. Qi, "Pedestrian detection in thermal images using adaptive fuzzy C-means clustering and convolutional neural networks," in Machine Vision Applications (MVA), 2015 14th IAPR International Conference on, 2015, pp. 246-249: IEEE.
[10] J. Han and B. Bhanu, "Fusion of color and infrared video for moving human detection," Pattern Recognition, vol. 40, no. 6, pp. 1771-1784, 2007.
[11] D. Xu, W. Ouyang, E. Ricci, X. Wang, and N. Sebe, "Learning cross-modal deep representations for robust pedestrian detection," arXiv preprint arXiv:1704.02431, 2017.
[12] D. Guan, T. Ma, W. Yuan, Y.-K. Lee, and A. Jehad Sarkar, "Review of sensor-based activity recognition systems," IETE Technical Review, vol. 28, no. 5, pp. 418-433, 2011.
[13] S. Iwasawa, K. Ebihara, J. Ohya, and S. Morishima, "Real-time estimation of human body posture from monocular thermal images," in Computer Vision and Pattern Recognition, 1997. Proceedings., 1997 IEEE Computer Society Conference on, 1997, pp. 15-20: IEEE.
[14] S. Maity, D. Bhattacharjee, and A. Chakrabarti, "A Novel Approach for Human Action Recognition from Silhouette Images," IETE Journal of Research, vol. 63, no. 2, pp. 160-171, 2017.
[15] X. Zhen and L. Shao, "A local descriptor based on Laplacian pyramid coding for action recognition," Pattern Recognition Letters, vol. 34, no. 15, pp. 1899-1905, 2013.
[16] K. Schindler and L. Van Gool, "Action snippets: How many frames does human action recognition require?," in Computer Vision and Pattern Recognition, 2008. CVPR 2008. IEEE Conference on, 2008, pp. 1-8: IEEE.
[17] R. Poppe, "A survey on vision-based human action recognition," Image and vision computing, vol. 28, no. 6, pp. 976-990, 2010.
[18] S. Vishwakarma and A. Agrawal, "A survey on activity recognition and behavior understanding in video surveillance," The Visual Computer, vol. 29, no. 10, pp. 983-1009, 2013.
[19] A. F. Bobick and J. W. Davis, "The recognition of human movement using temporal templates," IEEE Transactions on pattern analysis and machine intelligence, vol. 23, no. 3, pp. 257-267, 2001.
[20] J. Han and B. Bhanu, "Human activity recognition in thermal infrared imagery," in Computer Vision and Pattern Recognition-Workshops, 2005. CVPR Workshops. IEEE Computer Society Conference on, 2005, pp. 17-17: IEEE.
[21] T. H. Lam and R. S. Lee, "A new representation for human gait recognition: Motion silhouettes image (MSI)," in International Conference on Biometrics, 2006, pp. 612-618: Springer.
[22] H. A. Song and S.-Y. Lee, "Hierarchical representation using NMF," in International Conference on Neural Information Processing, 2013, pp. 466-473: Springer.
[23] Y. LeCun, Y. Bengio, and G. Hinton, "Deep learning," Nature, vol. 521, no. 7553, pp. 436-444, 2015.
[24] Y. LeCun et al., "Backpropagation applied to handwritten zip code recognition," Neural computation, vol. 1, no. 4, pp. 541-551, 1989.
[25] A. Krizhevsky, I. Sutskever, and G. E. Hinton, "Imagenet classification with deep convolutional neural networks," in Advances in neural information processing systems, 2012, pp. 1097-1105.
[26] C. Szegedy et al., "Going deeper with convolutions," in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2015, pp. 1-9.
[27] K. Simonyan and A. Zisserman, "Very deep convolutional networks for large-scale image recognition," arXiv preprint arXiv:1409.1556, 2014.
[28] A. Krizhevsky, I. Sutskever, and G. E. Hinton, "Imagenet classification with deep convolutional neural networks," in Advances in neural information processing systems, 2012, pp. 1097-1105.
[29] K. He, X. Zhang, S. Ren, and J. Sun, "Deep residual learning for image recognition," in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016, pp. 770-778.
[30] J.-Y. Bouguet, "Pyramidal implementation of the affine lucas kanade feature tracker description of the algorithm," Intel Corporation, vol. 5, no. 1-10, p. 4, 2001.
[31] M. Welling, M. Rosen-Zvi, and G. E. Hinton, "Exponential family harmoniums with an application to information retrieval," in Advances in neural information processing systems, 2005, pp. 1481-1488.
[32] P. Janik and T. Lobos, "Automated classification of power-quality disturbances using SVM and RBF networks," IEEE Transactions on Power Delivery, vol. 21, no. 3, pp. 1663-1669, 2006.
[33] I. Haritaoglu, D. Harwood, and L. S. Davis, "W/sup 4: real-time surveillance of people and their activities," IEEE Transactions on pattern analysis and machine intelligence, vol. 22, no. 8, pp. 809-830, 2000.
[34] R. K. Srivastava, K. Greff, and J. Schmidhuber, "Highway networks," arXiv preprint arXiv:1505.00387, 2015.
[35] Srivastava, Nitish, et al. "Dropout: A simple way to prevent neural networks from overfitting." The Journal of Machine Learning Research 15.1, 1929-1958, 2014.
[36] Kingma, Diederik, and Jimmy Ba. "Adam: A method for stochastic optimization." arXiv preprint arXiv:1412.6980, 2014.