Uncertainty analysis in the representation of the urban land cover classes through the application of artificial neural network

• Autor(a) principal: Boschi, Letícia Sabo
• Data de Publicação: 2007
• Outros Autores: Galo, Maria de Lourdes Bueno Trindade
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Boletim de Ciências Geodésicas
• Texto Completo: https://revistas.ufpr.br/bcg/article/view/8243
• topic: Classificação de ambientes urbanos, Redes Neurais Artificiais, Incerteza na classificação, Sensoriamento Remoto; Classification of urban environment, Artificial Neural Networks, Uncertainty in the classification, Remote Sensing

The great diversity of materials that characterizes the urban environment determinesa structure of mixed classes in a classification of multiespectral images. In thatsense, it is important to define an appropriate classification system using a nonparametric classifier, that allows incorporating non spectral (such as texture) data tothe process. They also allow analyzing the uncertainty associated to each class fromthe output values of the network calculated in relation to each class. Consideringthese properties, an experiment was carried out. This experiment consisted in theapplication of an Artificial Neural Network aiming at the classification of the urbanland cover of Presidente Prudente and the analysis of the uncertainty in therepresentation of the mapped thematic classes. The results showed that it is possibleto discriminate the variations in the urban land cover through the application of anArtificial Neural Network. It was also possible to visualize the spatial variation ofthe uncertainty in the attribution of classes of urban land cover from the generatedrepresentations. The class characterized by a defined pattern as intermediary relatedto the impermeability of the urban soil presented larger ambiguity degree and,therefore, larger mixture.Keywords: Classification of urban environment, Artificial Neural Networks,Uncertainty in the classification, Remote Sensing.