Abstract
The stabilizing functional approach is utilized to restore a distorted signal, the Itakura-Saito distortion measure of communication theory and the Kullback-Leilber distance of statistics are employed as the stabilizing functionals. The minimization of these nonquadratic functionals, subject to some statistical knowledge about the noise corrupting the system, is regarded as the minimization of a distortion measure or directed distance from a prior guess P to the desired signal z. The extremum of the unconstrained problem is then attained through an iterative scheme that allows for the introduction of two novel robust and stable signal restoration algorithms.< >