植基於雷射人體掃描辨識技術之跌倒偵測系統

The Falling Detect System Based on Human Laser Scanned Recognition Technique

洪榮裔(Jung-Yi Hung)

指導教授 ： 曾煥雯

國立臺灣師範大學/科技學院/應用電子科技學系/碩士(2012年)

摘要

近年來，因老人人口比例增加，老人的安養問題漸受各國重視，而研究發現，老人的日常生活中，遭遇到的意外以跌倒居多，而跌倒正是造成老人嚴重傷害的主因之一，因此如何在跌倒事件發生的當下立即發現並處理，便是當前社會的一個重要議題。
本研究提出一種以雷射掃描影像為基礎的人體跌倒偵測系統。整個系統架構分為建置階段、跌倒偵測階段與辨識階段，以震動型加速度計或是近接型光電感測器輔助系統運作。在人體跌倒姿勢的特徵萃取部份則包含了人體區域分割的演算法，及基於人體上緣包絡線的特徵萃取演算法。本文也提出一個效能的分析指標，以探討本系統的運作效能以及可改進之處。
藉由本文提出的研究方法、步驟以及流程，並經由實驗驗證，整體系統的辨識率，在辨識跪趴、撐趴、上躺與下躺等姿勢時，辨識率為79.55%；在辨識趴、坐與躺等三個姿勢時，辨識率為83.03%；在辨識站立的姿勢時，其辨識率可達到100%。

關鍵字

雷射掃描影像 ； 人體高低比例 ； 專家系統 ； 曲線擬合

並列摘要

In the times of population aging, the issues of caring the aged are becoming popular. According to foreign research, the most common accident in the daily life of the aged is falling down, which is the main cause of elder’s injury. As the result, how to detect and deal with the falling down event of the aged immediately is a big issue in the current society.
In this research, we proposed a human falling down gesture detecting system based on laser range image. The whole system is divided into building stages, falling detect stage, and recognition stage. The system use accelerometer and photoelectric sensors to help it run, and use the human body extraction algorithm and the feature extraction algorithm based on the edge line of the human body to extract the falling down gesture feature. Finally, it use the expert system to do the recognition work. In order to determine the efficiency of the system, we propose a performance indicator..
The approach, steps and flows proposed in this thesis via our experiments prove that the system can successfully recognize kneel-lie, prop-lie, up-lie and down-lie to the rate of 64.39%. When recognizing the lie, sit and lying down gestures, the success rate of the system is 78.79%. And when recognizing the stand gestures, the success rate of the system is 100%.

並列關鍵字

laser range image ； high-low ratio of human body ； expert system ； curve fitting

參考文獻

[2] A. Sixsmith and N. Johnson, “A Smart Sensor to Detect the Falls of the Elderly”, IEEE Pervasive computing, Vol.3, Issue 2, pp. 42-47, April, 2004.

[5] M. Alwan, P. J. Rajendran, S. Kell, D. Mack, S. Dalal, M. Wolfe, and R. Felder, “A Smart and Passive Floor-Vibration Based Fall Detector for Elderly”, 2nd IEEE International Conference on Information and Communication Technologies, Syria, Damasus, April 24-28, 2006.

[8] C. Doukas, L. Maglogiannis, P. Tragas, D. Liapis, and G. Yovanof, “Patient Fall Detction using Support Vector Machines”, Proceedings of International Federation for Information Processing, pp. 147-156, July, 2007.

[9] Q. Li, J. A. Stankovic, M. Hanson, A. Barth, and J. Lach, “Accurate, Fast Fall Detection Using Gyroscopes and Accelerometer-Derived Posture Information”, Body Sensor Network 2009, Berkeley, CA, June, 2009.

[10] M. Lan, A. Nahapetian, A. Vahdatpour, and L. Au, “SmartFall: An Automatic Fall Detection System Based on Subsequence Matching for the SmartCane”, 4th International Conference on Body Area Network (BodyNets), Los Angeles, California, April, 2009.