On the estimation of two-dimensional moving average parameters

The authors address the estimation problem of moving average (MA) parameters of a 2D autoregressive moving average (ARMA) model. The problem is equivalent to solving a set of overdetermined 2D transcendental equations. Based on some extensions of the Newton-Raphson method, an iterative algorithm is proposed for estimating 2D MA parameters. The performance of the algorithm is demonstrated by a numerical example. For 2D sinusoids in white noise spatial series, some interesting features of 2D ARMA modeling are observed.<
>