An adaptive time-frequency filtering algorithm for multi-component LFM signals based on generalized S-transform

Recent studies show that Cohen class bilinear time-frequency distribution methods do not have satisfactory denoising performance when analyzing multi-component LFM signals. This paper has constructed a new adaptive time-frequency filtering factor and has proposed an adaptive time-frequency filtering algorithm based on generalized S-transform. Firstly, the time-frequency distribution is obtained by transforming the time domain signals to time-frequency domain by using generalized S-transform, which is followed by calculating instantaneous frequency based on the phase information from the time-frequency distribution. Secondly, the time-frequency distribution regions occupied by clustered energy of effective signal are identified through time-frequency region extraction method and all time-frequency distribution spectrum out of the regions are removed. Thirdly, a novel TF filtering factor is constructed by the time-frequency concentration characteristic to restrain the random noise components in the regions of effective signal. Finally, the filtered signals are retrieved by using inverse generalized S-transform. Simulation results demonstrate that the proposed filtering algorithm has satisfactory performances for signal denoising which most features of original signal can be remained.