Abstract:Since reflection coefficients obey the generalized Gaussian distribution in most cases, the Lp(0.7< p< 1.3) norm can be chosen as a measurement of sparsity. And the generalized Gaussian distribution keeps weak reflection information better than the modified Cauchy distribution when p< 1. Based on the assumption of the reflection-coefficient generalized Gaussian distribution, a non-convex Lp norm regularized blind deconvolution model is build according to the Bayesian sparse inversion theory. True solutions of reflection coefficients and wavelets can be simultaneously obtained with the alternative relaxation strategy. In order to get stable solutions, the non-convex Lp(0< p< 1) norm regularized problem is transformed to a re-weighted least-squares problem. Numerical tests show that the proposed approach can extract wavelets and reflection coefficients and efficiently improve the vertical resolution.
Robinson E, Treitel S. Principles of digital Wiener filtering.Geophysical Prospecting,1967,15(3): 311-332.
[3]
Ulrych T J. Application of homomorphic deconvolution to seismology. Geophysics, 1971, 36(4): 650-660.
[4]
Wiggins R. Minimum entropy deconvolution. Geoexploration, 1978,16(1): 21-35.
[5]
Levy S, Oldenburg D. Automatic phase correction of common midpoint stacked data.Geophysics,1987,52(1): 51-59.
[6]
Wiggins R. Entropy guided deconvolution. Geophy-sics, 1985, 50(12): 2720-2726.
[7]
Kazemi N, Sacchi M D. Sparse multichannel blind deconvolution.Geophysics,2014,79(5):V143-V152.
[8]
Canadas G. A mathematical framework for blind deconvolution inverse problems. SEG Technical Program Expanded Abstracts, 2002, 21:2202-2205.
[9]
Taleb A, Sole J, Cadals I et al. Quasi-nonparametric blind inversion of Wiener systems. IEEE Transactions on Signal Processing, 2001, 49(5): 917-924.
[10]
Larue A,Van Der Baan M, Mars J et al. Sparsity or whiteness: what criterion to use for blind deconvolution of seismic data? SEG Technical Program Expanded Abstracts, 2005,24:1642-1646.
[11]
Sacchi M D.Statistical and Transform Methods for Seismic Signal Processing[D].University of Alberta, Edmonton, Canada,1999.
Li Haishan,Wu Guochen,Yin Xingyao.Seismic blind deconvolution method based on generalized Gaussian distribution.Progress in Geophysics,2012, 27(3): 936-944.
Li Guofa,Qin Dehai,Peng Gengxin et al.Experimental analysis and application of sparsity constrained deconvolution. Applied Geophysics,2013,10(2): 191-200.
[17]
Gray W C. Variable Norm Deconvolution[D].Stanford University,Stanford,California,USA,1979.
Cao Jingjie,Wang Yanfei,Yang Changchun.Seismic data restoration based on compressive sensing using the regularization and zero-norm sparse optimization.Chinese Journal of Geophysics,2012,55(2):596-607.