Skip to Main Content
This paper presents a novel and effective method for robust human tracking applied to far-infrared image sequences. It makes use of the characteristics of human body regions in far-infrared images and is based on a particle filter framework. The method constructs the regions of interestpsilas (ROI) histogram representation in an intensity-distance projection space, so as to hurdle the disadvantage of insufficient information when only intensity feature is considered. Furthermore, it correctly updates the above mentioned human body representation model which is embedded in the particle filter framework and propagate sample distributions over time. Experimental results using different far-infrared image sequences show the proposed scheme achieves more robust and stable than the classical tracking method.