An adaptive regularized method for deconvolution of signals with edges by convex projections

A new adaptive deconvolution method based on the projection operators onto convex sets (POCS) is presented. A minimum norm least-squares (MNLS) is obtained for signals with edges by means of an estimation-detection-protection scheme. The regularized differentiation technique is necessary for a reasonable detection of the signal edges. The improvement introduced with this method is illustrated through a simulation example. Finally, a discussion of the wide series of possibilities open along these lines closes this article.<
>