SEISMIC RANDOM NOISE ATTENUATION USING DIRECTIONAL TOTAL VARIATION IN THE SHEARLET DOMAIN

摘要

In this paper we propose an effective seismic denoising method using directional total variation (DTV) in the shearlet domain. This approach exploits the sparseness of shearlet transform and direction sensitivity of DTV. Shearlet shrinkage has a positive effect on denoising, but suffers from Gibbs artifact which can be solved by total variation (TV). DTV is an improved method of TV using anisotropic projection and performs well for the noisy signal with a dominant direction. The seismic data can be decomposed into several subbands by shearlet transform. Every subband has its own dominant direction. Therefore applying the direction information to DTV can more effectively eliminate seismic noise. The application on synthetic data and field data shows that the proposed method is superior to shearlet transform or DTV in seismogram noise removal and feature preservation.