Research On The Reconstruction,Denoising And Compression Of Seismic Signal In Wave Atoms Transform Domain

Seismic signal reconstruction,denoising and compression are both important parts of seismic signal processing,which are of great significance for seismic signal interpretation and storage.Due to the limited collection environment and exploration costs,the acquired seismic signals are often incomplete and contain various noises.In addition,with the development of petroleum exploration technology,the seismic signals acquired have become increasingly large.The wave atom is a kind of wavelet packet variant.Compared with wavelet,Gabor atom and curvelet,the wave atom has better sparse representation of the oscillating texture signal.And the wave atoms transform is a multi-scale geometric analysis method proposed in recent years.In this paper,the reconstruction,denoising and compression of seismic signals are carried out in the framework of the wave atoms transformation domain.The main contents are as follows:(1)Research on BCR iterative reconstruction of 3D seismic signal based on wave atoms transform.When seismic signals are collected in the field,there is usually a case of missing signals.It is necessary to reconstruct the missing track of seismic signals.Aiming at the three-dimensional(3D)missing seismic signal,this paper presents a Block Coordinate Relaxation(BCR)iterative reconstruction algorithm based on wave atoms transform.The BCR algorithm is used to iteratively reconstruct the seismic signal under the framework of wave atoms transform.Firstly,3D wave atoms transform is done on the 3D missing seismic signal,and then the hard threshold function with exponential attenuation is utilized to deal with the transformed coefficients.Finally the reconstruction result is obtained by inverse wave atoms transform.The reconstruction has conducted on 3D synthetic and the North Sea F3 field seismic signal with random sampling experiments,and compared with wave atoms,curvelet,wavelet and dual-tree complex wavelet transform based algorithms.The reconstructed results demonstrate that the proposed algorithm outperforms other three reconstruction algorithms.(2)Research on blind denoising of 3D seismic signal in the wave atoms transform domain.In view of the fact that the seismic data will inevitably be mixed with noise during the sampling process,and the noise level is unknown.In this paper,a blind denoising algorithm for 3D seismic signals in wave atoms transform domain is constructed.The main idea is that noise estimation algorithm is used to estimate the amount of noise in the seismic signal.Then wave atoms transformation is used for the seismic signal after cycle spinning processing.Using the estimated noise parameters,the thresholds are selected at different scales,and then correct them.A new improved threshold function is used to deal with the wave atoms coefficients,and then the inverse wave atoms transform and the inverse cycle spinning are performed.Finally,the denoised result of 3D seismic signal is obtained.The synthetic and real seismic signal with noise are denoised and compared with the denoising results of wavelet transform,Dual-tree complex wavelet transform,curvelet transform and traditional wave atoms transform.The results show that the proposed algorithm has obvious advantages over other compared algorithm.With the increase of noise in seismic signal,the denoising advantage becomes more obvious.From the evaluation indicators including output signal to noise ratio,mean square error and peak signal to noise ratio can be analyzed,the best denoising algorithm is the blind denoising of 3D seismic signal in wave atom transform domain,followed by the traditional wave atoms transform algorithm,then the curvelet transform algorithm and dual-tree complex wavelet transform algorithm,the wavelet transform algorithm perform worst.(3)Seismic data compression based on embedded zerotree wave atoms.In order to improve storage efficiency and reduce transmission time and cost,each link of seismic signal processing require high-quality compression algorithms.In this paper,a signal compression algorithm based on embedded zero tree wave atom(EZWA)is proposed and used to compress seismic signals.The basic idea of EZWA algorithm is to obtain the coefficients of the signal by the wave atoms transformation firstly,then encode the wave atoms coefficients and quantify the encoded coefficients,and finally get the encoded compressed signals.In order to verify the effectiveness of the EZWA algorithm,the compression has conducted on Marmousi model seismic signal,field marine pre-stack seismic signal and field land post stack seismic signal experiments,and compared with the traditional EZW compression algorithm.Under the premise of the same compression ratio,it can be seen from the index of peak signal to noise ratio,mean square error and mean absolute error that the compression and reconstruction of EZWA algorithm is better than EZW.And the EZWA algorithm has a higher compression ratio under the similar peak signal-to-noise ratio,which shows that the EZWA algorithm can get better compression than EZW.