Small object monitoring for sensitive indoor compounds

Abstract

This paper describes the development of an automatic human monitoring system for indoor environments. The aim was to develop a system that allows general movement of one or more people to be determined using a very low-resolution colour sensor. The sensor has sufficiently low resolution that objects are not resolved – to protect privacy. It is demonstrated that movement can be monitored to a high degree of accuracy using spectral un-mixing techniques adapted from satellite remote sensing applications. These algorithms allow the fractional contributions from different colours within each pixel to be estimated and this can then be used to assist in the detection and tracking of small objects. Of particular interest is the robustness of the monitoring algorithms in a series of real-life challenging scenarios such as quickly changing illumination conditions and occlusion of objects