Generalized Hessian-Schatten norm regularization for image reconstruction

Regularization plays a crucial role in reliably utilizing imaging systems for scientific and medical investigations. It helps to stabilize the process of computationally undoing any degradation caused by physical limitations process. In past decades, total variation regularization played dominant literature. Two forms regularizations, namely first-order second-order (TV-1 TV-2) have been widely used. TV-1 has disadvantage: it reconstructs images form piece-wise constants when noise and/or under-sampling is severe, while TV-2 natural-looking under such scenarios. On other hand, can recover sharp jumps better than TV-2. generalizations, Hessian-Schatten norm (HSN) regularization, generalized (TGV) proposed become significant developments area inverse problems owing their performance. While strength TGV that combine advantages TV-2, HSN structure-preserving property. Here, we develop novel image recovery combines strengths HSN. We achieve this restricting maximization space dual same way obtained from call new (GHSN). computational method reconstruction using based on well-known framework called alternating direction multipliers (ADMM). demonstrate GHSN some examples.