Community Research and Development Information Service - CORDIS


The classification of remotely-sensed satellite imagery for thematic mapping applications can be improved by the use of ancillary geographic information derived from maps. Two methodologies have been developed for using such information in the automatic processing of imagery: an expert system method and a neural network method. Both approaches require accurate co-registered datasets derived from a GIS. In this work, ARC/INFO has been used to process the map data and to make it available to the classifiers along with the imagery. Initial results have shown that the ancillary GIS data contribute towards greater accuracy in the image product and, in the neural network case, lead to much faster training of the classifier.

Additional information

Authors: WILKINSON G G, JRC Ispra (IT);BURRILL A, JRC Ispra (IT)
Bibliographic Reference: Paper presented: EGIS/European Conference on Geographical Information Systems, Bruxelles (BE), April 2-5, 1991
Availability: Available from (1) as Paper EN 35991 ORA
Record Number: 199110428 / Last updated on: 1994-12-02
Original language: en
Available languages: en