Seismic imaging profile noise suppression based on self supervised deep learning: a case study in the Yellow Sea Available to Purchase

Seismic data imaging profiles are often affected by migration noise. This article uses a deep learning denoising method using a self coding convolutional neural network to suppress spatial aliasing and random noise in seismic imaging stacked profiles. This article designs an autoencoder model, mainly composed of convolutional layers, pooling layers, dropout layers, and upsampling layers, to automatically learn and extract features from seismic data. Through the encoding and decoding structure, the goal of noise attenuation on seismic imaging profiles is achieved. The test results on forward simulation data and actual seismic data in the Yellow Sea region show that the proposed method has good suppression effects on spatial aliasing and random noise.