Nonlocal Total Variation Denoising of Seismic Data

Abstract

Seismic denoising can be considered to be a total variation minimization problem. Nonlocal total variation (NLTV) denoising is one of the best denoising models and is widely used in image processing. Combined with Split-Bregman algorithm, the computational efficiency of NLTV regularization can be improved, making it able to handle large data set. We propose to adopt the NLTV regularization method to attenuate random noise of seismic data and compare it with f-x deconvolution. Experimental results show that the NLTV denoising performs better.