A new technique for order determination of two-dimensional arma models

In this paper, a new method for order determination of two-dimensional autoregressive moving-average (2-D ARMA) models is proposed. In the method proposed, the ARMA model order is dependent on the minimum eigenvalues of a covariance matrix derived from the observed data. The 2-D ARMA model is assumed to be causal, stable, linear, and spatial shift-invariant with p1 x p2 quarter plane (QP) support. Numerical simulations are presented to illustrate the importance of this new approach in model order determination.