Random noise attenuation is one of the critical task for seismic data pre-conditioning. Especially, suppression of strong noise has been attracting increasing attention. In this abstract, the theory of simultaneously sparse coding is introduced to strong random noise suppression in seismic data. Combined with the low-rank based algorithm named SAIST, nonlocal means method is applied to extract an initial estimate from each cluster before solving the optimization problem. Furthermore, a designed filter which calculates the lateral coherence of vector sets in each cluster is used to determine the extracting process. Synthetic test on model corrupted by strong noise shows that better de-noising performance can be obtained by the proposed algorithm. Application of field seismic records also verifies the practical perspective of this method.
Presentation Date: Monday, September 16, 2019
Session Start Time: 1:50 PM
Presentation Start Time: 1:50 PM
Location: Poster Station 13
Presentation Type: Poster
