The Research Of Suppressing Random Noise Method Based On General S Transform

Time-frequency analysis is a powerful tool for nonstationary time-varying signal analysis, and becomes a hot spot in modern signal processing research. Traditional time-frequency analysis methods are summarized in this paper, including the origin of time-frequency analysis, the classification of time-frequency analysis, and gives the commonly used form of time-frequency analysis, for example, short-time Fourier transform, Wigner-Ville distribution, Wavelet transform, S transform. S transform is a method of time-frequency analysis developed in recently years, and soon to be applied to seismic data processing and interpretation. In this paper we highlight on the origin, nature and characteristics of S transform, and introduce the generalized S transform.We compare the effects of time-frequency analysis of the short time Fourier transform, S transform and Generalized S transform by the theoretical model, conclusions is high frequency components of signal of short-time Fourier transform and S transform can not corresponding reflection coefficient of thin, not suitable for identification and estimate the thickness of thin bed. Compare the filtering effect of Fourier transform and S transform, conclusions is time-frequency domain filtering by general S transform to suppressing noise effect better than the frequency domain filtering by Fourier transform , reduce the parameter k in the general S transform filtering effect can be improved. Using Generalized S transform to suppress random noise of model and actual CMP gather and achieving good results.