Random noise in seismic signal is a kind of interference wave,it seriously reduces the quality and signal-to-noise ratio of seismic signal,and affects the subsequent processing and data analysis.Therefore,it is particularly important to denoise for seismic signal.Seismic signal noise reduction is refers to,according to the actual earthquake field collected signal to estimate the original seismic signal,namely in reducing noise and better retain the actual effective information of seismic signals.Fractional order B-splines wavelet has optional order time,can be more flexible to adjust the wavelet coefficients;Gaussian scale mixture model can well describe the marginal distribution of wavelet transform coefficient and the relationship between the coefficient of neighborhood.In this thesis,the fractional order B-splines wavelet transform combined with Gaussian scale mixture model is used to suppress random noise in seismic signal.The main research work of this thesis are as follows:1.Summarizes the development of seismic signal noise reduction,this thesis expounds the fractional order composition and properties of B-splines wavelet,and Gaussian scale mixture model,this thesis introduces the evaluation standard of seismic signal noise reduction.2.The fractional order B-splines wavelet transform combined with Gaussian scale mixture model for seismic signal to suppress random noise.Using fractional order B-splines wavelet transform seismic signals with noise signal is mapped to the optimal fractional wavelet time-frequency domain,and then to the wavelet subband coefficients Gaussian scale mixture model respectively,by the Bayesian method to estimate the source seismic signal wavelet coefficient,finally using fractional order B-splines wavelet noise reduction is obtained by inverse transformation to reconstruct the seismic signal.Based on the synthetic seismograph and actual seismic signal noise reduction processing,shows that the method can effectively suppress the random noise in seismic signal,better retain the signal effectively. |