People Tracking in Ambient Assisted Living Environments Using Low-Cost Thermal Image Cameras

摘要

In this paper we propose the use of low-cost thermal imaging sensors for the application of people tracking in Ambient Assisted Living environments. We describe background subtraction and segmentation on low-resolution thermal images as the necessary preprocessing steps to derive suitable percepts. Extensive data association to samples of a set of Monte-Carlo particle filters is realized by circle-circle intersection tests between percepts and pose hypotheses. Experimental evaluation conducted in the Bremen Ambient Assisted Living Lab proves the precision of the proposed system by comparing the tracking results against ground truth data coming from an ARTTRACK optical tracking system.