Home // AMBIENT 2015, The Fifth International Conference on Ambient Computing, Applications, Services and Technologies // View article

Monitoring Abnormal Behavior of Hospital Patients Using RGB+D Sensors

Authors:
Seung Chae
Kin Choong Yow

Keywords: Abnormal behavior detection, learning, Kinect, data mining, Ambient Assisted Living(AAL)

Abstract:
The ability to recognize abnormal actions or conditions of hospital patients is a very important problem as it may bring about timely medical response to a critical patient condition and may make the difference between life and death. In this paper, we propose a system that makes use of the Microsoft Kinect for Windows v2 to generate RGB+D (Red, Green, and Blue + Depth) image sequences of hospital patients. Dense 2D image features were extracted from the image sequences and then combined in a hierarchical manner to form compound features. These compound features were then mined to produce a class feature model to be used for action recognition. In the recognition phase, the RGB and the depth image data were processed separately and the responses merged to produce an overall response for action classification. Our experimental results show that this approach is able to generate good recognition rates and is comparable to other state of the art algorithms.

Pages: 10 to 15

Copyright: Copyright (c) IARIA, 2015

Publication date: July 19, 2015

Published in: conference

ISSN: 2326-9324

ISBN: 978-1-61208-421-3

Location: Nice, France

Dates: from July 19, 2015 to July 24, 2015