Research On Seismic Data Reconstruction Method Based On Compressed Sensing

As the environment of seismic exploration becomes more and more complex,the precision of seismic imaging becomes higher and higher.Reconstruction of high-resolution seismic data has always been a hot topic in the field of seismic exploration.How to balance SNR and resolution has always been a difficult problem in seismic imaging.Traditional seismic data processing methods have been difficult to meet people's requirements for fine characterization of complex geological structures.Therefore,this paper studies the seismic data reconstruction method based on compressed sensing.Based on the two theoretical frameworks of compressed sensing and sparse representation,this paper makes in-depth study and research on two aspects of high resolution reconstruction of seismic data: denoising and frequency extension reconstruction.The main contents are as follows:In this paper,the theoretical basis of compressed sensing and sparse representation is first studied and re-understood.Two design requirements of measurement matrix,restricted isometry property and correlation analysis,are described in detail.By comparing several sparse representation methods,the corresponding advantages and disadvantages are listed,and the principle of sparse representation method based on dictionary learning is elaborated.In the aspect of seismic data denoising,the method of seismic denoising based on K-SVD dictionary learning is studied.According to the characteristics of seismic data itself,this method can be adaptive to train the optimal dictionary for sparse representation of seismic data,so as to realize the separation of noise and effective signal and achieve the purpose of denoising.In the selection of reconstruction algorithm,the orthogonal matching tracking algorithm is improved to control the number of iterations by expected parameters,and the influence of expected parameters on the denoising effect is analyzed.Finally,the theoretical model and practical data are compared with some traditional denoising methods to verify the effectiveness of the algorithm.For the reconstruction of seismic data,a method based on wavelet modulus maxima sparse extension frequency reconstruction is proposed in this paper.The sparsity selection of mother wavelet and the theory of wavelet modulus maxima are analyzed.Combining with the propagation characteristics of wavelet modulus maxima,the measurement matrix is designed.The wavelet modulus maxima of each scale are compressed and sampled,and then the wavelet coefficients of each scale are compressed and sensed.The reconstruction has been verified by theoretical model and practical data,and good results have been achieved.