Two-dimensional lattice linear prediction parameter estimation method and fast algorithm

Abstract

This paper presents an extension of the one-dimensional (1-D) lattice (reflection coefficient) technique of linear prediction parameter estimation, first popularized by Burg, to the two-dimensional (2-D) case. The resulting fast recursive 2-D algorithm is a significant computational simplification over and an estimation improvement on previous attempts to extend the 1-D Burg linear prediction algorithm to 2-D by exploiting some newly discovered matrix structures. The technique presented here is useful for high resolution 2-D spectral analysis applications and the creation of high-resolution spotlight-mode synthetic aperture radar (SAR) imagery, as is illustrated.