A stopping rule for iterative image restoration with constraints

A method is presented for estimating the optimal stopping point in a constrained iterative image restoration algorithm. Terminating an iterative algorithm early can improve the restoration considerably. The optimal number of iterations cannot be determined a priori. The optimal number of iterations changes when constraints are enforced in the iterative procedure. The generalized cross-validation criterion is modified to estimate the optimal stopping point in the presence of constraints on the solution. An efficient estimation procedure is adopted that allows evaluation of the criterion in an iterative manner with a computational complexity on the same order as the constrained restoration. The benefit of this procedure is demonstrated with examples