The implementation of self organised neural networks for cloud classification in digital satellite images

Author

Stephanidis, C.N. ; Cracknell, A.P. ; Hayes, L.W.B.

Author_Institution

Dept. of Appl. Phys. & Electron. & Manuf. Eng., Dundee Univ., UK

Volume

1

fYear

34881

fDate

10-14 Jul1995

Firstpage

455

Abstract

Presents the results of a study where a self organised map (SOM) was used to classify a NOAA-AVHRR satellite image. The neural network was fed with both spectral and spatial features. All five bands were used to extract the median and range of small “floating” subwindows as well as the average entropy and the range of average entropy. This procedure resulted in a vector with four components per band for each pixel. These vectors were processed by the neural network to obtain a projection into a two-dimensional Kohonen map

Keywords

atmospheric techniques; clouds; geophysical signal processing; geophysics computing; image classification; image colour analysis; meteorology; optical information processing; self-organising feature maps; AVHRR; IR visible infrared; NOAA; atmosphere meteorology; average entropy; cloud; cloud classification; digital satellite image; floating subwindow; image classification; image colour analysis; measurement technique; multispectral imaging; neural net; optical imaging; remote sensing; satellite image; self organised map; self organised neural network; spatial feature; two-dimensional Kohonen map; Clouds; Entropy; Image storage; Intelligent networks; Mechanical engineering; Neural networks; Neurofeedback; Neurons; Physics; Satellite broadcasting;

fLanguage

English

Publisher

ieee

Conference_Titel

Geoscience and Remote Sensing Symposium, 1995. IGARSS '95. 'Quantitative Remote Sensing for Science and Applications', International

Conference_Location

Firenze

Print_ISBN

0-7803-2567-2

Type

conf

DOI

10.1109/IGARSS.1995.520307

Filename

520307