Target-oriented least-squares migration/inversion with sparseness constraints

conclusions

I have presented a regularized least-squares inversion scheme to image the reflectivity. This inversion scheme allows us to perform inversion in a target-oriented fashion, and the total cost is about two migrations (one for computing the migrated image, the other for computing the phase-encoded Hessian). Examples on the Marmousi model show that regularization that promotes sparsity in the image domain help to reduce the null space and to mitigate the effects of operator mismatch. Inversion with the sparseness constraint can lead to a better solution with higher resolution than that regularized with the standard $\ell _2$-norm damping.