Classification of remotely-sensed satellite images using multi-layer perceptron networks
Multilayer feed-forward perceptron networks have been used to classify high resolution remotely-sensed satellite data for land cover mapping. Experiments on a test area in southern France using multitemporal multispectral images from the SPOT-1 satellite have demonstrated the potential of neural networks for discriminating large numbers of ground cover classes. An average classification accuracy of the order of 84% has been obtained for 20 classes.
Bibliographic Reference: Paper presented: International Conference on Artificial Neural Networks (ICANN '91), Espoo (FI), June 24-28, 1991
Availability: Available from (1) as Paper EN 36042 ORA
Record Number: 199110804 / Last updated on: 1994-12-02
Original language: en
Available languages: en