New fading memory fast SRLS algorithm for 2-D SAR model parameter estimation

Author

Zhao, Ping Ya ; Litva, John

Author_Institution

Commun. Res. Lab., McMaster Univ., Hamilton, Ont., Canada

fYear

1993

fDate

14-17 Sep 1993

Firstpage

700

Abstract

We present a new fast spatially recursive least-squares (SRLS) algorithm with exponentially fading memory for adaptive estimation of two-dimensional (2-D) nonstationary simultaneous autoregressive (SAR) model parameters. The computational complexity of the new algorithm is 8m^3/2+6m multiplications and divisions per recursion (MADPR) in contrast with 15m^3/2+16m MADPR of the best existing algorithm, where m is the number of the estimated model parameters. The new algorithm has the same statistical properties and tracking capability, compared with the existing algorithms. The derivation of the algorithm and the computer simulation results are given in the paper

Keywords

computational complexity; estimation theory; image processing; least squares approximations; parameter estimation; stochastic processes; time series; 2-D SAR model parameter estimation; adaptive estimation; computational complexity; computer simulation results; digital image processing; exponentially fading memory; fast SRLS algorithm; multidimensional system identification; nonstationary simultaneous autoregressive model; spatially recursive least-squares; statistical properties; tracking; Adaptive estimation; Additive white noise; Computational complexity; Computer simulation; Fading; Laboratories; Parameter estimation; Recursive estimation; Strontium; Two dimensional displays;

fLanguage

English

Publisher

ieee

Conference_Titel

Electrical and Computer Engineering, 1993. Canadian Conference on

Conference_Location

Vancouver, BC

Print_ISBN

0-7803-2416-1

Type

conf

DOI

10.1109/CCECE.1993.332392

Filename

332392