Community Research and Development Information Service - CORDIS


In this paper the use of multi-sensor and multi-temporal data for land cover classification is investigated. In particular a multilayer Neural Network (NN) trained by means of the Back Propagation algorithm is employed for classification experiments on remotely sensed images.
The data set employed is composed of two coregistered images of the agricultural area surrounding the city of Valladolid (Spain) acquired in two different dates, June and July, by two different satellites, SPOT and LANDSAT respectively. Ground truth data was acquired during an in situ campaign carried out by the Spanish Ministry of Agriculture in 1993.
For the experiments presented here, three different data sets were employed: SPOT data; LANDSAT data; and (SPOT + LANDSAT) data. In both second and third cases, the LANDSAT image was resampled in order to have the same resolution as the SPOT image.
In this paper the use of this integrated data set by means of a Neural Network is investigated and the results of land cover classification on the three different data sets outlined above are compared and discussed. d.

Additional information

Authors: CHIUDERI A, JRC Ispra (IT)
Bibliographic Reference: Paper presented: Time-varying image processing and moving object recognition, Florence (IT) September 5-6, 1996
Availability: Available from (1) as Paper EN 40071 ORA
Record Number: 199611140 / Last updated on: 1996-10-28
Original language: en
Available languages: en