Post-FWI Injection of Learned Priors Using a Flow Matching Model
Full Waveform Inversion (FWI) is a powerful tool for subsurface velocity reconstruction but remains highly ill-posed, sensitive to acquisition limitations, often requiring some form of regularization to reduce artifacts and enhance resolution. While recent developments have shown that generative models can inject learn...