Uma arquitetura neural modular para classificação de imagens multiespectrais de sensoriamento remoto

AUTOR(ES)
DATA DE PUBLICAÇÃO

1997

RESUMO

This work presents an Artificial Neural Network (ANN) based architecture for the classification of Remote Sensing (RS) multispectral imagery. This architecture consists of two processing modules: an Image Feature Extraction Module using Kohonen s Self-Organizing Map (SOM) and a Classification Module using a Multi-Layer Perceptron (MLP) network. This architecture was developed aiming at two specific goals: to exploit the advantages of unsupervised learning for feature extraction and the testing of techniques to increase the learning algorithms performance concerning training time. More specifically, this work tests the implementation of parallel learning algorithms for Kohonen s SOM in a multiprocessing environment and the utilization of a second-order learning algorithm for the MLP network. The experimental results exhibit a much superior performance by both algorithms. To test the applicability of this work, this architecture was applied to the classification of a LANDSAT/TM image segment from a pre-selected testing area and its performance was compared with that of a Maximum Likelihood Classifier. The good results obtained by the neural classification allied to the performance improvement encourage therefore further research efforts in this area

ASSUNTO(S)

sensoriamento remoto processamento de imagens - tecnicas digitais redes neurais (computação) inteligencia artificial

ACESSO AO ARTIGO

Documentos Relacionados