Decision-directed adaptive image restoration using multiple image and blur models
A. Tanju Erdem, Ahmet Murat Tekalp · 2005
We present an adaptive image restoration algorithm that is capable of restoring space-variant, blurs and of suppressing ringing artifacts simultaneously. Suppression of ringing artifacts is achieved by using multiple image models that provide a better match to local edge orientation. Space-variant blur restoration is done by using multiple blur models, assuming that the space-variant blur can be represented by a finite set of blur models. A reduced update Kalman filtering procedure with image-model and blur model detection based on maximum a posteriori probability (MAP) decision is used to restore an image.