Strip Kalman filtering for image restoration: new results

A new procedure for parameter identification of the image model in strip Kalman filtering is suggested. This method, which takes into account the vector nature of the image model, leads to a vector Yule-Walker system of equations. An algorithm for online adaptation of the image model parameters is also given. In the composite dynamic model for the image and degradation processes, the blur states are computed one step ahead of those of the image. Thus, a new set of Kalman filtering equations that accounts for both one-step prediction for the blur states and filtering for the image states is given.<
>