Through-wall tracking using mean-shift algorithm

Abstract

In through-wall-radar-imaging tracking, the target, such as a human body, is better to be regarded as a nonrigid cluster rather than a point target, because the target image usually occupies a lot of pixels in the image and dramatically changes its profile over time. Thus, instead of the traditional point tracking algorithm, we try to use the mean shift algorithm to deal with the problem, as it has been proved to be suitable for real-time visual tracking of nonrigid objects. Before mean shift algorithm, the strong clutter, such as the multi-path ghosts and grating-lobe artifacts, is suppressed by the phase coherence factor (PCF) to provide high signal-to-clutter-ratio images. Simulation and experimental results show that the mean shift algorithm works very well in through-wall moving target tracking. Furthermore, the mean shift algorithm has a very low computational complexity which makes the real-time tracking possible.