Group Sparse Optimization for Images Recovery Using Capped Folded Concave Functions

This paper considers the image recovery problem by taking group sparsity into account as prior knowledge. is formulated a sparse optimization over intersection of polyhedron and possibly degenerate ellipsoid. It convexly constrained with cardinality objective function. We use capped folded concave function to approximate show that solution set continuous approximation solutions are same. Moreover, we penalty method replace constraints in adding convex nonsmooth existence positive parameters such sets unconstrained propose smoothing algorithm any accumulation point sequence generated directional stationary problem. Numerical experiments for presented illustrate efficiency adaptive functions.