Comparison of radar-based human detection techniques

Radar offers unique advantages over other sensors in the human detection problem, such as remote operation during virtually all weather and lighting conditions. Many radar-based human detection systems today employ Fourier analysis, such as spectrograms. However, spectrograms perform poorly in high clutter environments. Also, an inherent SNR loss is caused by the implicit assumption of linear target phase. In this paper, human modeling is used to derive a more accurate non-linear approximation to the true non-linear target phase and the liklihood ratio is optimized over unknown parameters to enhance detection performance. Performance Is compared both analytically and through MATLAB simulations.