Denoising of UAV Micro-Doppler Signatures Based on BEMD and Wavelet Transform

In recent years, the detection and classification of UAVs has become a hot issue. The difficulty is that UAVs have very small RCS and usually fly at low altitudes. Radar echo contains strong noise, clutter and other interference. To solve this problem, this paper proposes a denoising algorithm combining bivariate empirical mode decomposition(BEMD) and wavelet transform based on UAV micro-Doppler characteristics. First, the pre-processed radar signal is decomposed into complex modal components by BEMD decomposition. Then the peak value of the autocorrelation function is used to automatically select the IMF components containing the micro-Doppler feature to perform wavelet threshold denoising, and finally the micro-Doppler signal is reconstructed. Experiments were carried out using simulation and measured data of UAV rotor blades. Experimental results show that the proposed method has a significant denoising effect at low Signal-Noise-Ratio(SNR).