Application of Improved Anisotropic Diffusion Filtering in Seismic Data Denoising

Zaiping Nie,Hong Wang

DOI: https://doi.org/10.3997/2214-4609.202310629

2023-01-01

Abstract:Summary In seismic exploration, noise generated by many factors affects the quality of seismic image, and further affects the subsequent seismic data processing. Conventional filtering methods such as Gauss filtering, median filtering, and F-K domain filtering etc usually blur the useful structural information of seismic data while denoising. Anisotropic filtering can remove random noise and protect structural features in seismic images, so it is widely used in seismic data processing. We propose an improved diffusion coefficient method for anisotropic filtering. The diffusion coefficient of the improved method depends not only on the local energy of the image, but also on the local dip angle of the image. The intensity of diffusion filtering is constrained by the local dip of the structure. When there are faults or large-dip structures in seismic images, the dip constraint makes the filtering coefficient small and the diffusion filtering intensity weak. In the flat region with good continuity or small-dip structures, the filtering coefficient is large and the diffusion filtering intensity is large. Compared with the traditional anisotropic diffusion equation denoising, the proposed method can keep the large-dip structure from being destroyed in the filtering. The improved method obtains good results from synthesized data and real data.