Posture Recognition Based Fall Detection System
Adel Rhuma, Miao Yu, Jonathon A. Chambers · Lecture Notes on Software Engineering · 2013
In this paper, we introduce a video-based fall detection system for an elderly person in a realhome environment. We extract global (ellipse) and local (shape context) features from static postures and an improved Directed Acyclic Graphic Support Vector Machine (DAGSVM) is applied for posture classification. After classifying different postures, certain rules are set to detect falls. This fall detection system is shown by evaluation on real datasets to achieve a good fall detection performance.