Geoprocessamento na discriminação de pastagens degradadas utilizando rede neural artificial em imagem Ikonos II

• Ano de defesa: 2006
• Autor(a) principal: Silva, Edgley Pereira da
• Orientador(a): Cantarutti, Reinaldo Bertola
• Banca de defesa: Gleriani, José Marinaldo , Demattê, José Alexandre Melo
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Viçosa
• Programa de Pós-Graduação: Doutorado em Solos e Nutrição de Plantas
• Departamento: Fertilidade do solo e nutrição de plantas; Gênese, Morfologia e Classificação, Mineralogia, Química,
• País: BR
• Palavras-chave em Português: Geoprocessamento; Pastagens; Degradação; Rede neural artificial; Sistemas de informação geográfica; Sensoriamento remoto
• Palavras-chave em Inglês: Geoprocessing; Pastures; Environmental degradation; Artificial neural network; Geographic information systems; Remote sensing
• Área do conhecimento CNPq: CNPQ::CIENCIAS AGRARIAS::AGRONOMIA::CIENCIA DO SOLO
• Link de acesso: http://locus.ufv.br/handle/123456789/1671

The cattle in Brazil reached from 60s, an increasing expansion and consequently an increase in areas of pastures. However, the gradual loss of the productive capacity of these pastures after some years has been a constant, mainly in extensive areas of exploration. The recovery of degraded pastures is hard to implement due to lack of information brought up to date and detailed regarding its space distribution. The use of the remote sensing, a technology whose data can be gotten to the low cost, with repetitivity and in a compatible scale with the dimension of problem, can highly contribute for the solution of this problem. The area of study is located at the city of João Pinheiro, in the northwest of the state of Minas Gerais, enclosing an area of 100 km2. The aims of this study were: a) the diagnosis and the quantification of the area according to the systems of use and the stadium of productivity of pastures and b) the application of methods of automatic classification in high-resolution imagery using Artificial Neural Networks. Image of the satellite Ikonos II with space resolution of one meter was used. The exactness of the mapping was verified using the Kappa index. The visual interpretation of the image and field-work made possible the identification of pastures in the following categories of productivity: productive, low degradation, average degraded, and degraded. The present pastures in the area were predominantly of Brachiaria brizantha (1,693 ha), Andropogon gayanus (192 ha) and Panicum maximum (231 ha). In accordance with the Kappa index, the visual classification of image Ikonos II has high potential to diagnosis the productivity of the pastures. The techniques of classification in images Ikonos II had revealed efficient in the discrimination of the levels of degradation of the pastures. The methods of maximum likelihood and ANN had not differed statistically; therefore, they present high potential for classification of high-resolution imagery. The qualitative evaluation of these images showed a satisfactory result. The method ISODATA was statistically inferior to the high-resolution imagery. With the reached results, new perspectives for future works confide, using ANN and different situations, contributing to the development of the digital image processing.