Imaging Domain Seismic Denoising Based on Conditional Generative Adversarial Networks (CGANs)

A high-resolution seismic image is the key factor for helping geophysicists and geologists to recognize geological structures below subsurface. More more complex geology has challenged traditional techniques resulted in a need powerful denoising methodologies. The deep learning technique shown its effectiveness many different types of tasks. In this work, we used conditional generative adversarial network (CGAN), which special type neural network, conduct process. We considered task as an image-to-image translation problem, transfers raw with multiple noise into reflectivity-like without noise. several models train CGAN. experiment, CGAN’s performance was promising. trained CGAN could maintain structure undistorted while suppressing