Estimation and identification for 2-D block Kalman filtering

Abstract

This correspondence is concerned with the development of a recursive identification and estimation procedure for 2-D block Kalman filtering. The recursive identification scheme can be used on-line to update the image model parameters at each iteration based upon the local statistics within a block of the observed noisy image. The covariance matrix of the driving noise can also be estimated at each iteration of this algorithm. A recursive procedure is given for computing the parameters of the higher order models. Simulation results are also provided.