|Least-squares diffraction imaging using shaping regularization by anisotropic smoothing|
We have developed an efficient approach to highlight and denoise edge diffractions based on least-squares migration. The inverted operator corresponds to the chain of path-summation integral filter, AzPWD and Kirchhoff modeling operators. While the combination of path-summation integral filter and AzPWD emphasizes edge diffraction signatures in the data domain, thresholding and anisotropic smoothing precondition them in the model domain by denoising and enhancing their continuity. Both forward modeling and shaping regularization operators guide the inversion towards restoration of edge diffractions. Synthetic and field data examples show high fidelity of the approach.
The efficiency of the proposed inversion scheme comes from the workflow application in time post-stack domain and shaping regularization framework leading to fast convergence. The inversion scheme we propose can be thought of as an effective operator directly tailoring edge diffractions and extracting them from the full wavefield.