Minimization of non-smooth, non-convex functionals by iterative thresholding (Q2346390): Difference between revisions

The paper proposes a new approach (CGPM) to the minimization of non-smooth, non-convex functionals, which involves a family of proximity mappings for non-convex functionals. The key is to compute iteratively the inverse of a given proximity mapping. The authors first define this generalized gradient projection method, and then later they study in a detailed manner one-dimensional penalty functions, which are then applied to separable constraints. The special case when the penalty function is the power function is considered in a detailed manner and some numerical examples are displayed.