Parameter estimation for two-dimensional vector models using neural networks

Abstract

This correspondence addresses the problem of two-dimensional (2-D) vector image model parameter estimation using a new recursive least squares (RLS)-based learning method. Vector autoregressive (AR) models with various 1-D and 2-D, causal and noncausal regions of support (ROS) are considered. Numerical results are presented which demonstrate the usefulness of the proposed scheme for on-line implementation.