Denoising Of Seismic Signal Based On Block Matching And Cooperative Filtering

Seismic exploration is an important measure for finding oil and gas.But the field seismic signals which actual collected often have random noise interference,which seriously affects the imaging quality and interpretation accuracy of seismic signals.Based on the non-local mean and sparse representation theory,this paper studies the seismic signal denoising method based on block matching collaborative filtering algorithm.The main research work in this paper is as follows:(1)Denoising of two-dimensional seismic signal combined curvelet transform and BM3 D.This paper proposes a novel seismic data reconstruction signal denoising algorithm combined curvelet transform and BM3 D.On the one hand,the basic idea is the algorithm preprocesses the seismic signal with the curvelet domain hard-threshold denoising,and integrates the denoising result into the basic estimation part of BM3 D.On the other hand,the noisy seismic signal is input directly to the final estimation part of BM3 D for subsequent wiener filtering and block estimation aggregation.The algorithm not only improves the accuracy of the BM3 D basis estimation,but also provides the energy spectrum closer to the original signal for the Wiener filter in the final estimation part of BM3 D.Theoretical and real seismic signal denoising results show that this algorithm can effectively denoise,and the proposed algorithm has a better denoising effect compared with the simple curvelet domain denoising and BM3 D denoising.(2)Based on noise estimation of principal component analysis BM3 D seismic signal denoise.BM3D denoising need to predict the noise intensity,but in practical applications,the noise intensity of the seismic signal is unknown.This paper presents a BM3 D two-dimensional seismic signal denoising algorithm combined with noise estimation of principal component analysis.The basic idea is to block the noisy seismic signals firstly,Then,by calculating data blocks the smallest eigenvalue of covariance matrix to estimate the noise variance.And the noise standard deviation of the estimated value input to the BM3 D,to complete the seismic signal denoising.Theoretical and real seismic signals of the experimental results illustrates that the proposed method of this paper is superior to other similar algorithms in accuracy of noise estimation,and has good denoising effect on seismic signals.(3)Three-dimensional seismic signal denoising based on block matching and collaborative filtering.In this paper,the block matching four-dimensional cooperative filtering(BM4D)algorithm is applied to three-dimensional seismic signal denoising.Considering that wavelet transform can not capture the singularity of seismic signals.Therefore,the original BM4 D algorithm based on the bior1.5 wavelet transform are changed to discrete cosine transform to improve the basis of the estimated part of the signal to noise ratio.Give a closer energy spectrum of the original signal to experience wiener filtering of BM4 D final estimate.The experiments on synthetic and actual three-dimensional seismic signals illustrates that the proposed method can denoising three-dimensional seismic signals very well,comparing with BM4 D denoising algorithm based on bior1.5,bior2.4 and Dmey wavelet transform.