Total generalized variation based denoising models for ultrasound images (Q2399157)

For the restoration of noisy ultrasound images, the authors present minimization models based on convex/nonconvex total generalized variation (TGV) regularization. To realize the convex TGV model, the alternating direction method of multipliers is used. A convergence analysis for the convex TGV model is given, too. For the nonconvex TGV model, the iteratively reweighted \(\ell_1\) algorithm is applied. This model preserves the edges of the restored images well. Both methods result in fast and efficient optimization algorithms. Numerical experiments demonstrate the performance of these methods.