Covariance estimation with limited training samples

Author

Tadjudin, Saldju ; Landgrebe, David A.

Author_Institution

Sch. of Electr. & Comput. Eng., Purdue Univ., West Lafayette, IN, USA

Volume

37

Issue

4

fYear

1999

fDate

7/1/1999 12:00:00 AM

Firstpage

2113

Lastpage

2118

Abstract

This paper describes a covariance estimator formulated under an empirical Bayesian setting to mitigate the problem of limited training samples in the Gaussian maximum likelihood (ML) classification for remote sensing. The most suitable covariance mixture is selected by maximizing the average leave-one-out log likelihood. Experimental results using AVIRIS data are presented

Keywords

Bayes methods; covariance analysis; geophysical signal processing; geophysical techniques; image classification; remote sensing; terrain mapping; AVIRIS; Bayes method; Gaussian maximum likelihood classification; average leave-one-out log likelihood; covariance estimation; covariance estimator; covariance mixture; empirical Bayesian setting; geophysical measurement technique; image classification; image processing; land surface; limited training sample; optical imaging; remote sensing; terrain mapping; Bayesian methods; Covariance matrix; Degradation; Density functional theory; Maximum likelihood estimation; Military computing; NASA; Parameter estimation; Remote sensing;

fLanguage

English

Journal_Title

Geoscience and Remote Sensing, IEEE Transactions on

Publisher

ieee

ISSN

0196-2892

Type

jour

DOI

10.1109/36.774728

Filename

774728