Adaptive Image Reconstruction Using Information Measures

We present a class of nonlinear adaptive image restoration filters which may be steered to preserve sharp edges and contrasts in the restorations. From a theoretical point of view we discuss the associated variational problems and prove existence of solutions in certain Sobolev spaces W 1,p or in a BV -space. The degree of regularity of the solution may be understood as a mathematical explanation of the heuristic properties of the designed filters.