Application And Research Of Random Seismic Noise Attenuation Based On The CWT

The SNR of seismic data has been a key point in seismic exploration. Due to itscorrelation, multi-resolution characteristic, wavelet transform is widely used inde-noising of seismic exploration.Although the method of wavelet threshold has beenused widely in the production due to its simple and practical characteristics, hardthreshold function are not continuous and it can lead the signal to producepseudo-Gibbs phenomenon, there is always a certain degree of deviation between theestimated value and the actual value.With exploration environment becoming increasingly diverse, exploration targetstransfer more and more to the deep, conventional wavelet methods cannot meet thehigh requirement of SNR of the seismic data. The threshold has been a key factor inthe process of de-noising, this paper give the basic concepts, ideas and conclusions ofwavelet transform, including continuous wavelet transform theory, dyadic wavelet,wavelet frame theory, multiresolution analysis and Mallat algorithm, describedseveral classical wavelet transform de-noising algorithm in detail and derived thenoise estimate using the regression method., then discussed the theory and propertiesof complex wavelet in detail, since the complex wavelet transform has thecharacteristics of approximate translation invariance, good directional selectivity,complete refactoring features and limited redundancy, so it can overcome the existingburr of discrete wavelet transform.Conventional wavelet threshold cannot fully use the characteristics of the seismicsignals due to its own limitation, for this, this paper give two binary shrink functionde-noising models (BivaShrink13, BivaShrink23) starting from the originalBivaShrink12model and made complex wavelet improvements to them, finally weuse the current coefficient, the parent layer coefficient, and the neighborhoodcoefficient, taking into account the correlation of inter-layer and inner layer wavelet coefficients based on DTCWT and related sub-band de-noising model (TrivaShrink),to achieve the optimal estimation of shrinking factor of the wavelet function and toget the noise reduction of seismic records. Through model test, we prove that complexwavelet transform is better than conventional wavelet methods in removing randomnoise of seismic records.