RECURSIVE ESTIMATION OF IMAGES USING NON-GAUSSIAN AUTOREGRESSIVE MODELS

We consider recursive estimation of images modeled by non-Gaussian aut oregressive (AR) models and corrupted by spatially white Gaussian nois e. The goal is to find a recursive algorithm to compute a near minimum mean square error (MMSE) estimate of each pixel of the scene using a fixed lookahead of D rows and D columns of the observations. Our metho d is based on a simple approximation that makes possible the developme nt of a useful suboptimal nonlinear estimator. The algorithm is first developed for a non-Gaussian AR time-series and then generalized to tw o dimensions, In the process, we draw on the well-known reduced update Kalman filter (KF) technique of Woods and Radewan [1] to circumvent c omputational load problems, Several examples demonstrate the non-Gauss ian nature of residuals for AR image models and that our algorithm com pares favorably with the Kalman filtering techniques in such cases.