ANALYSIS OF REGULARIZED TOTAL VARIATION PENALTY METHODS FOR DENOISING

The problem of recovering images with sharp edges by total variation d enoising has recently received considerable attention in image process ing. Numerical difficulties in implementing this nonlinear filter tech nique are partly due to the fact that it involves the stable evaluatio ns of unbounded operators. To overcome that difficulty we propose to a pproximate the evaluation of the unbounded operator by a stable approx imation. A convergence analysis for this regularized approach is prese nted.