Morphological component analysis for feature detection in satellite images

Author

Koren, Ilan ; Joseph, Joachim H.

Author_Institution

Res. Council - NRC, NASA Goddard Space Flight Center, Greenbelt, MD, USA

fYear

2003

fDate

27-28 Oct. 2003

Firstpage

70

Lastpage

72

Abstract

A new approach to cluster analysis is proposed, namely morphological component analysis (MCA), to enhance discrimination of features in multi-channel satellite images. The characterization of clusters, in this method, is morphological, unlike some of the classical cluster approaches in which the clusters are defined by their centers. By using the shape and orientation of the clusters, it is possible to define an affine transformation of the cluster space into a new one in which the selected clusters are orthogonal or better separated. Such an operation can be considered as supervised independent component analysis.

Keywords

feature extraction; image classification; independent component analysis; pattern clustering; remote sensing; affine transformation; cluster analysis; feature detection; independent component analysis; morphological component analysis; multichannel satellite images; Clouds; Computer vision; Councils; Image analysis; Image color analysis; Independent component analysis; Oceans; Satellites; Shape; Storms;

fLanguage

English

Publisher

ieee

Conference_Titel

Advances in Techniques for Analysis of Remotely Sensed Data, 2003 IEEE Workshop on

Print_ISBN

0-7803-8350-8

Type

conf

DOI

10.1109/WARSD.2003.1295174

Filename

1295174