📝 SummarXiv

Abstract

Seismic data acquisition is often affected by various types of noise, which degrade data quality and hinder subsequent interpretation. Recovery of seismic data becomes particularly challenging in the presence of strong noise, which significantly impacts both data accuracy and geological analysis. This study proposes a novel single-encoder, multiple-decoder network based on Nash equalization (SEMD-Nash) for effective strong noise attenuation in seismic data. The main contributions of this method are as follows: First, we design a shared encoder-multi-decoder architecture, where an improved encoder extracts key features from the noisy data, and three parallel decoders reconstruct the denoised seismic signal from different perspectives. Second, we develop a multi-objective optimization system that integrates three loss functions-Mean Squared Error (MSE), Perceived Loss, and Structural Similarity Index (SSIM)-to ensure effective signal reconstruction, high-order feature preservation, and structural integrity. Third, we introduce the Nash Equalization Weight Optimizer, which dynamically adjusts the weights of the loss functions, balancing the optimization objectives to improve the models robustness and generalization. Experimental results demonstrate that the proposed method effectively suppresses strong noise while preserving the geological characteristics of the seismic data.