使用多攝影機作人群異常行為偵測

Detection of Abnormal Behavior within A Crowd of People Using Multiple Cameras

周昱呈(Yu-Chen Chou)

指導教授 ： 莊仁輝

國立交通大學/資訊學院/多媒體工程研究所/碩士(2017年)

摘要

隨著自動監視系統的普遍，在公共區域中可以看到許多監視攝影機存在，因此也越來越多研究注重在即時人物定位，應用於避免一些重大災難發生。然而在公共區域等多人環境中，因為路人相碰撞後跌倒，或是在偵測範圍中有人感到身體不適，進而跌倒時，此時即時偵測到人物發生跌倒之情況，便能即時進行搶救。本論文中推廣先前一種針對在偵測範圍之中進行即時人物定位方法，以進一步分析偵測到的人是否發生跌倒情況。該方法採用加速基於垂直線消失點的線段取樣方法，是一種能快速人物定位的方法，能夠定位人群，並且在相當短的時間內精確地估計其高度，推廣的做法是利用高度進行人物行為分析，將人物行為分成跌倒與非跌倒之兩種情況。除了可以利用人物高度的變化進行跌倒偵測，也能利用於其他的異常行為，例如：小朋友進入社區的危險區域。

關鍵字

人物定位 ； 異常行為偵測 ； 多攝影機 ； 身高分析

並列摘要

With the popularity of automatic surveillance system, many surveillance cameras can be seen in public areas. Recently, more and more research works focus on real-time people localization and pose detection system to avoid some major catastrophes. For example, if the system can promptly detect that people are falling down in the monitored area, due to pushing or a health problem, proper rescue actions may be taken to prevent the problem turning into uncontrollable situations. In this paper, we extend a previous real-time people localization method to further analyze the event of people falling. The method uses a line sampling scheme based on the vanishing point of vertical lines to achieve fast people localization for a group of people, while the extended method uses estimated human height to classify human behavior into the two cases of fall and non-fall. The height of the person changes not only can be used to detect the fall, but also can be used for other abnormal behavior, for example, children enter the community of dangerous areas.

並列關鍵字

People localization ； Detection of Abnormal Behavior ； Multiple Cameras ； Height analysis

參考文獻

[1] H. Nait-Charif and S.J. McKenna, “Activity Summarisation and Fall Detection in a Sup-portive Home Environment,” in Proceedings of the 17th International Conference on Pat-tern Recognition, Aug. 2004.

[2] A. Palaniappan, R. Bhargavi, V. Vaidehi, “Abnormal Human Activity Recognition Using SVM Based Approach,” International Conference on Recent Trends in Information Tech-nology, pp. 97-102, April 2012.

[3] E. Akagündüz, M. Aslan, A. Şengür, H. Wang, and M. C. İnce, “Silhouette Orientation Volumes for Efficient Fall Detection in Depth Videos,” IEEE Journal of Biomedical and Health Informatics, Vol. 21, No. 3, pp 756-763, May 2017.

[4] F. Merrouche and N. Baha, “Depth Camera Based Fall Detection Using Human Shape and Movement,” IEEE International Conference on Signal and Image Processing (IC-SIP), pp. 586-590, Aug. 2016.

[5] E. E. Stone and M. Skubic, “Fall Detection in Homes of Older Adults Using the Microsoft Kinect,” IEEE Journal of Biomedical and Health Informatics, Vol. 19, No. 1, pp. 290-301, Jan. 2015.