Date of Award

2012

Degree Type

Thesis

Degree Name

Bachelor of Computer Science (Honours)

School

School of Computer and Security Science

Faculty

Computing, Health and Science

First Advisor

Dr Martin Masek

Second Advisor

Associate Professor C. Peng Lam

Abstract

The growing elderly population presents a challenge on the resources of carers and assisted living communities. This has led to various projects in remote automated home monitoring in order to keep the elderly in their own home environment longer. These have the promise of alleviating the strain on support services, and the benefits of keeping people in their existing familiar community environment. Such monitoring typically involves a myriad of sensors attached to the environment and person, so as to acquire rich enough data to determine the actions of the person being monitored. In this research, an algorithm based around the Microsoft Kinect, its 3D camera and person detection features, is presented for monitoring activities of daily living. The Kinect was originally released as a device to improve human interaction in gaming for the Xbox 360 game console. In this approach, training Data is obtained and then preprocessed using Kinect’s in build person recognition. Collection of training data is necessary is because the process in which Activities of Daily Living (ADLs) are completed varies from person to person, and therefore no generic template to recognise ADLs exists. The research attempts to infer representations of ADLs through features related to the spatial position of the person in the field of view of the Kinect camera, which is divided into a grid of several data points. Once this representation or “ADL signature” is obtained, ADLs are classified live in unknown data using a combination of distance measures, and abnormal events are detected amongst these ADLs using an automatically generated threshold value. This system has the potential to replace the various sensors in the camera’s field of view, and provide a system that accurately analyses the behaviours of the elderly to provide doctors and carers with valuable observational data.

Share

 
COinS