Another stopping rule for linear iterative signal restoration

A new stopping rule is proposed for linear, iterative signal restoration using the gradient descent and conjugate gradient algorithms. The stopping rule attempts to minimize MSE under the assumption that the signal arises from a white noise process. This assumption is appropriate for many coherent imaging applications. The stopping rule is trivial to compute and, for flxed relaxation parameters, can be computed prior to starting the iteration. The utility of the stopping rule is demonstrated through the restoration of MR imagery.