Community Research and Development Information Service - CORDIS

Abstract

In the framework of the application of Remote Sensing techniques for the selective inventory of natural resources (poplar groves in the Po Valley) the goal of the present paper is to examine the possibility to extrapolate, over large areas, classification methods defined and tested over few rather small regions. In the first part we analyse the effects, related to the use of different learning regions, both on the spectral definition of the categories to be recognized and on the classification results. In the second part, in order to improve the performances of these computer-aided classification methods, we propose and discuss a new iterative technique (coupled supervised and unsupervized feature selection) and, finally, we present the results obtained using a completely unsupervized classification scheme.

Additional information

Authors: BEONIO BROCCHIERI F JRC ISPRA ESTAB. (ITALY), JRC ISPRA ESTAB. (ITALY)
Bibliographic Reference: EUR 9109 IT (1984) MF, 51 P., BFR 120, BLOW-UP COPY BFR 255, EUROFFICE, LUXEMBOURG, POB 1003
Availability: Can be ordered online
Record Number: 1989122098600 / Last updated on: 1987-01-01
Category: PUBLICATION
Available languages: it
Follow us on: RSS Facebook Twitter YouTube Managed by the EU Publications Office Top