Community Research and Development Information Service - CORDIS

Abstract

The requirements for obtaining, rapidly, precise estimations of crop acreages throughout the European Communities, have necessitated the development of a machine-assisted photointerpretation technique, which makes use of an expert system. Its implementation aims to enhance the first classification map, given by one of the conventional classifiers (Maximum Likelihood-Minimum Distance) and to obtain in a post-classification level process more realistic map products. The rule base of the expert system exhibits a considerable variety of information, which relates to the image and geographic context of the classified pixel. Probabilistic information from the image classifier and from the rule base is combined using the Dempster-Shafer model of evidential reasoning. An experiment is described in which soil maps, buffered road networks and a generalised classification map have been used as additional layers for classifying SPOT XS images from the Department Loir-et-Cher in France. Tests using ground data have shown that the use of the expert system gives an additional accuracy improvement of 12.8%.

Additional information

Authors: KONTOES C C, JRC Ispra (IT)
Bibliographic Reference: Paper presented: Application of Remote Sensing to Agricultural Statistics, Belgirate (IT), Nov. 26-27, 1991
Availability: Available from (1) as Paper EN 36551 ORA
Record Number: 199210468 / 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