Yuan, SY; Wang, SX.JOURNAL OF GEOPHYSICS AND ENGINEERING, 2013, 10(2): 025001发布时间：2013-09-27
Edge-preserving noise reduction based on Bayesian inversion with directional difference constraints
Yuan, SY; Wang, SX*
Attenuating random noise while preserving edges and structures of seismic signals is of great significance for seismic interpretation and inversion. An edge-preserving and signal-preserving noise reduction method is presented in this paper. The method regards noise reduction problem as an inverse problem based on a Bayesian framework. A priori information that the difference data of seismic signals obey Cauchy distribution, a kind of non-Gaussian distribution, is used as a constraint to control the noise reduction. The performance of the method is mainly dependent on a trade-off parameter and a scale parameter, which make it easier to obtain a relatively perfect signal preservation and noise reduction result. With directional constraints, the method recovers linear events and quasi-linear events without dip steering. In particular, the method can adaptively protect edges of discontinuous events. A synthetic data example and two field post-stack data sets illustrate the effectiveness of the method.