Human Motion Recognition Based on Radar Micro-Doppler Features Using Bayesian Network

Human target is a kind of widely concerned target, this paper aims at walk, run and jump three kinds of human activities to identify. In the simulation experiment, first of all, using the single frequency continuous-wave radar measures walk, run and jump three kinds of activities to obtain the corresponding echo signal. Secondly, the time-frequency analysis applied to echo signals, and the spectrograms of three kinds of action are obtained respectively. Then can extract the required micro-Doppler features from spectrograms. Finally, the human motion can be identified by using Bayesian network based on fusion micro-Doppler features.