深度影像之物體傾角計算暨在跌倒偵測之應用

Abstract

近年來隨著醫療技術的進步，國人平均壽命逐年提升，老年人口大幅增加，如何為老年人提供安全的居住環境已經成為一項重要的議題。而根據統計指出，跌倒造成的傷害為事故傷害中的第二大主因，不少老年人在發生跌倒後，因身體無法行動或呈現意識不清的狀態，導致無法及時尋求救援，進而可能造成更嚴重的傷害，因此，為了能在老年人發生跌倒意外時，能即時地獲得救援與協助，行人的跌倒偵測之研究應運而生。

本論文利用深度攝影機擷取深度資訊後，對行人計算3D傾角。經由本研究推導之最小平方法(LS)來計算行人3D傾角，並將3D傾角計算法實現於行人跌倒偵測系統上，最後驗證此方法在跌倒偵測系統上的可行性。

In the recent years, with the advances in medical technology, the elderly population have grown up substantially. How to provide a safety environments for the elderly has become an important issue. And according to the statics, the injuries caused by fall is one of the main reason in the accidental injuries. In addition, some elderly people would lose the mobility and even the conscious after the fall happened, which might delay the rescue time or lead to more serious damages. Therefore, in order to provide the immediately rescue when the accident fall occurs, so we proposed a novel method for pedestrian fall detection.

In our research, we adopt the depth-camera to capture the data of depth, and use the least-square method to calculate the tilt angle of pedestrian. Then, we could implement the method to the fall detection system. Finally, the effectiveness of the fall detection have been verified.