Two dimensional AR modeling by autocorrelation fitting

Document Type

Conference Proceeding

Date of Original Version



The problem of two-dimensional AR (autoregressive) modeling of two-dimensional stationary random fields by fitting the estimated autocorrelation function (ACF) with the model autocorrelation function is addressed. Both nonsymmetric half-plane (NSHP) and quarter-plane (QP) supports are used for modeling. Ekstrom-Woods two-dimensional spectral factorization is used to generate a stable and accurate initial estimate of the model coefficients and permits the determination of the model coefficients with the same support as the original autocorrelation data. Several experimental results are given to illustrate the effectiveness of the method in two-dimensional artificial random field synthesis and spectral estimation.

Publication Title, e.g., Journal

ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings



This document is currently not available here.