Abstract: Conventional generalized S transform (GST) uses fixed parameters for Gaussian window, which makes it impossible to take into account the high and low frequency signals in the time-frequency analysis. At the same time, the standard S inverse transform produces filtering noise when filtering in the time-frequency domain. In this paper, a generalized S transform based on variable frequency Gaussian window is proposed, and the inverse S transform formula is revised. This method not only improves the focus of the frequency spectrum of the signal, but also eliminates the filtering noise. By calculating the instantaneous signal-to-noise ratio threshold value of the signal containing random interference noise, and according to different threshold values, the processing strategy of suppressing the random noise is selected in a targeted manner. The processing results of synthetic data and actual seismic data show that the method can effectively suppress random noise and improve the signal-to-noise ratio of seismic data.