Community Research and Development Information Service - CORDIS

Abstract

Experiments have been conducted on the classification of remotely-sensed satellite imagery using multi-layer perceptron neural networks. The implementation of such networks yields results of high accuracy. Training of large networks required for realistic satellite mapping applications can require several hours of computer processing, although classification can be performed very rapidly. Some indications are given of techniques to design appropriate network architectures and to control the training procedures for fast convergence.

Additional information

Authors: KANELLOPOULOS I, JRC Ispra (IT);WILKINSON G G, JRC Ispra (IT);MÉGIER J, JRC Ispra (IT);VARFIS A, JRC Ispra (IT)
Bibliographic Reference: Paper presented: International Space Year Conference, München (DE), March 30 - April 4, 1992
Availability: Available from (1) as Paper EN 36931 ORA
Record Number: 199210949 / Last updated on: 1994-12-02
Category: PUBLICATION
Original language: en
Available languages: en
Follow us on: RSS Facebook Twitter YouTube Managed by the EU Publications Office Top