Automatic classification of RADARSAT SAR images of the Northern Sea Route

Author

Bodanov, A.V. ; Sandven, Stein ; Johannessen, Ola M. ; Alexandrov, Vitaly Yu

Author_Institution

Nansen Int. Environ. & Remote Sensing Center, St. Petersburg, Russia

Volume

2

fYear

1999

fDate

1999

Firstpage

1028

Abstract

Two automatic sea ice classification algorithms based on linear discriminant analysis (LDA) and feedforward backpropagation multilayer perceptron (MLP) were compared for classification of RADARSAT ScanSAR images of the Western part of the Northern Sea Route. Texture and local statistical parameters in addition to radar image brightness are used as input features to the algorithms. Simultaneous in-situ observations were carried out during an expedition to the Barents and Kara seas afforded to assess the performance of the algorithms

Keywords

backpropagation; feedforward neural nets; geophysical signal processing; geophysics computing; image classification; image texture; multilayer perceptrons; oceanographic regions; oceanographic techniques; remote sensing by radar; sea ice; spaceborne radar; synthetic aperture radar; Arctic Ocean; Barents Sea; Kara Sea; Northern Sea Route; RADARSAT SAR image; algorithm; automatic classification; feedforward backpropagation multi-layer perceptron; feedforward neural network; image classification; image texture; linear discriminant analysis; local statistical parameters; neural net; ocean; radar image brightness; radar imaging; radar remote sensing; sea ice; spaceborne radar; synthetic aperture radar; Backscatter; Brightness; Classification algorithms; Linear discriminant analysis; Multilayer perceptrons; Navigation; Radar imaging; Remote sensing; Rivers; Sea ice;

fLanguage

English

Publisher

ieee

Conference_Titel

Geoscience and Remote Sensing Symposium, 1999. IGARSS '99 Proceedings. IEEE 1999 International

Conference_Location

Hamburg

Print_ISBN

0-7803-5207-6

Type

conf

DOI

10.1109/IGARSS.1999.774522

Filename

774522