In the context of ageing societies, assessing risk factors and detecting falls for the elderly is becoming a crucial issue. In this paper, we propose an iterative head tracking method based on particle filtering using the fusion of low cost thermal and depth sensors for home environments whilst preserving privacy. The iteration process begins by segmenting the head in the depth image to calculate the depth coefficients and the thermal coefficients used for updating the particle weights. The method was tested on several sequences, with or without depth-thermal fusion: results show its robustness and accuracy , and also demonstrate that fusion improves tracking, namely when fast motion occurs (in case of a fall for instance) or when segmentation is erroneous.