Fast adaptive identification and restoration of images degraded by blur and noise

A fast efficient technique for adaptive restoration and identification of images corrupted by noise and causal or noncausal blurs is presented. The method is based on the appropriate interweaving of a fast multichannel adaptive scheme for parameter estimation and on Kalman filtering of an equivalent state-space formulation of the image. This formulation, which uses a slightly augmented state vector, permits image restoration to include deconvolution. A combined adaptive scheme is thus proposed for the simultaneous estimation of image model and blur parameters and restoration of the degraded image pixels. Experimental results which illustrate the performance of the method are given.<>