處於高齡化人口結構的臺灣,高齡者獨居的比例日益增高,因此高齡者的醫療照護問題也顯得比以往更加重要。其中跌倒為高齡者常發生的意外事件之一,跌倒意外事件的發生往往會對高齡者的生理及心理造成一定程度傷害,嚴重的跌倒意外事件甚至可能威脅到高齡者的生命安全,因此許多相關產業與研究學者紛紛投入跌倒偵測與預防之相關研究領域。 在許多的跌倒偵測研究中經常使用頭部的高度作為判斷跌倒動作的特徵,但僅使用頭部的高度作為跌倒判斷的特徵,往往容易將頭部位置較低的日常生活動作誤判為跌倒動作。此外,過去的跌倒偵測研究在人體動作取得上通常使用較昂貴的配戴式專業設備或以傳統的影像攝影機結合影像處理技術來取得人體動作特徵,但往往會有裝置昂貴、使用者對配戴裝置的不適應及影像處理程序繁瑣且耗時等問題。 本研究透過無標記動作擷取的Kinect感應器自動取得人體動作資訊,提出結合雙關節組合、雙軸座標、雙特徵判斷之複合跌倒判斷方法,進行跌倒動作之偵測。此方法可降低跌倒與其他類似跌倒的日常生活動作之誤判,以提高跌倒偵測的正確辨識率。實驗結果顯示,本研究提出的複合跌倒判斷方法,可有效的區別跌倒與頭部高度位置較低的蹲下及撿東西之動作,有效改善先前研究僅使用頭部高度作為跌倒判斷之特徵容易產生誤判之問題。未來期望將此跌倒偵測方法應用於實際的居家跌倒偵測,提供高齡者於居家照護上更好的保障。
Nowadays Taiwan is an aged society. The population of solitary aged people are increasingly. Therefore, how to take care the aged people becomes a main topic for today. The fall down accident usually happened in old people, it may physically and mentally hurt the old person. The serious fall down may even critically danger the old people’s lives. Therefore, to prevent and detect fall down of old people becomes a research issue. Many studies of fall detection use the height of head to determine the characteristics of fall down. However, it may make a mistake when people do their Activities of Daily Living (ADL) below the height of head. The past studies used expensive tools, equipment or traditional cameras with leveraging the professional image techniques to get the features of human actions. But it has problems such as spending too much on the equipment; users can not fit those equipment or feel uncomfortable when they are wearing it. It also takes too much time to process the images. This study uses the markless action tool Kinect to extract the information of human actions automatically. The proposed method includes three features: dual joints, dual coordinates, and dual features of fall. It may decrease the mistakes of false detection of ADL. The experiment result shows that the proposed method can successfully distinct the fall down and the ADLs. It also decreases the possibility of false detection. This study suggests to use the proposed method to detect fall accident at home in the future. It may take care of the aged people and provide them a better life.