《Explicit_algorithms_for_a_new_time_dependent_model_based_on_level_set_motion_for_nonlinear》.pdf

EXPLICIT ALGORITHMS FOR A NEW TIME DEPENDENT MODEL BASED ON LEVEL SET MOTION FOR NONLINEAR DEBLURRING AND NOISE REMOVAL ANTONIO MARQUINA y AND STANLEY OSHER Dedicated to the memory of Emad Fatemi Abstract In this paper we formulate a time dependent mo del to approximate the solution to the nonlinear total var iation optimization problem for deblurr ing and nois e removal intro duce d by Rudin and Osher and Rudin Osher and Fatemi re spectively Our mo del is bas e d on level s et motion whos e steady state is quickly reache d by means of an explicit pro ce dure bas e d on Roes scheme us e d in uid dynamics We show numer ical evidence of the spee d of re solution and stability of this simple explicit pro ce dure in some repre s entative D and D numer ical example s Intro duction The class ical algor ithms for image deblurr ing andor denoi s ing have b een mainly bas e d on least square s Four ier s er ie s and other L 2norm approxi mations and cons equently the ir outputs may b e contaminate d by Gibbs phenomena and do not approxim ate well image s containing e dge s The ir computational advantage come s f rom the f act that they are linear thus f ast solvers are widely available How ever the eect of the re storation i s not lo cal in spatial scale Other bas e s of orthogonal functions have b een intro duce d in order to get r id of thos e problems eg compactly supp orte d wavelets However Gibbs phenomenon ringing i s still pre s ent for the s e norms The Total Var iation TV deblurr ing and denoi s ing mo dels are bas e d on a var i ational problem with constraints us ing the total var iation norm as a nonlinear non dierentiable functional The formulation of the s e mo dels was rst gi