Calibration Method for Micro-Doppler Radar to Identify Pedestrian Posture

Micro-Doppler radar has the potential to facilitate the recognition of pedestrian posture. However, the current micro-Doppler radar calibration approaches have limited effectiveness. Accordingly, with prescribed walking posture feature parameters, a novel radar calibration method was proposed for the generation of micro-Doppler signals via a semi-physical simulation platform. The principles of the micro-Doppler effect were initially elucidated through an analysis of the translational and rotational motions of a rigid body. Subsequently, a pedestrian model was established by simplified the structure of the pedestrian into 17 interconnecting scattering points, which represent key parts of the pedestrian. Furthermore, we developed a SystemVue software-based system for calibrating the micro-Doppler radar. IQ modulator and fractional-order Fourier transform were also employed to extract and analyze the micro-Doppler signals generated by the human body model. Then, a micro-Doppler versus time plot for micro-Doppler signals was obtained. During the experiment, the relative error of the pedestrian velocity is no more than 3%, which demonstrates the reliability and effectiveness of the proposed method.