Using a new threshold function in seismic data de-noising

Abstract In geological exploration, there is a large amount of unpredictable noise interference in seismic data that needs to be removed. The traditional soft and hard threshold methods have been widely used in seismic data de-noising, but they have some defects such as discontinuity and constant deviation at the threshold, which leads to poor de-noising effects. Therefore, this paper introduces the weighted average idea into the traditional threshold methods and an improved threshold function based on seismic data de-noising is proposed. In the new method, this paper explores the layers of wavelet decomposition, establishes an optimal wavelet basis, and selects a variability threshold according to the number of decomposition layers instead of a fixed constant, avoiding the elimination of some effective signals. The de-noising effect of new method is evaluated by using two evaluation indicators: signal-to-noise ratio abbreviated as SNR and root mean square error abbreviated as RMSE. Finally, the designed new method is applied to actual mining areas for seismic data de-noising. The results show that the new method can effectively remove noise from seismic data and the texture of seismic profile after noise reduction is clear and smooth. Compared with traditional threshold methods, the new method has better de-noising performance.