Fast algorithm for the two-dimensional modified covariance method of linear prediction

Author

Marple, S. Lawrence, Jr.

Author_Institution

Orincon Corp., San Diego, CA, USA

Volume

2

fYear

1998

fDate

1-4 Nov. 1998

Firstpage

1452

Abstract

This paper presents the foundation for extending the one-dimensional (1-D) modified covariance method of linear prediction to the two-dimensional (2-D) data/image application. A fast computational algorithm for the solution of the least squares normal equations of the 2-D modified covariance method will be provided. The fast algorithm exploits the centrosymmetric and near-to-doubly-Toeplitz-plus-Hankel structure of the normal equations when expressed in matrix form. The algorithm is useful for applications such as generating high-resolution synthetic aperture radar images.

Keywords

Hankel matrices; Toeplitz matrices; covariance matrices; image resolution; prediction theory; radar imaging; radar resolution; spectral analysis; synthetic aperture radar; 1D modified covariance method; 2D modified covariance method; Toeplitz-plus-Hankel structure; centrosymmetric structure; data/image application; fast computational algorithm; high-resolution SAR radar image generation; least squares normal equations; linear prediction; spectral analysis; synthetic aperture radar; Covariance matrix; Equations; High-resolution imaging; Image sampling; Least squares methods; Mesh generation; Nonlinear filters; Spatial resolution; Two dimensional displays; Vectors;

fLanguage

English

Publisher

ieee

Conference_Titel

Signals, Systems & Computers, 1998. Conference Record of the Thirty-Second Asilomar Conference on

Conference_Location

Pacific Grove, CA, USA

ISSN

1058-6393

Print_ISBN

0-7803-5148-7

Type

conf

DOI

10.1109/ACSSC.1998.751567

Filename

751567