%0 Conference Proceedings
%4 sid.inpe.br/mtc-m16c/2017/12.01.19.18
%2 sid.inpe.br/mtc-m16c/2017/12.01.19.18.51
%@issn 2179-4820
%T Simultaneous multi-source and multi-temporal land cover classification using a Compound Maximum Likelihood classifier
%D 2017
%A Reis, Mariane Souza,
%A Dutra, Luciano Vieira,
%A Escada, Maria Isabel Sobral,
%@affiliation Instituto Nacional de Pesquisas Espaciais (INPE)
%@affiliation Instituto Nacional de Pesquisas Espaciais (INPE)
%@affiliation Instituto Nacional de Pesquisas Espaciais (INPE)
%E Davis Jr., Clodoveu A. (UFMG),
%E Queiroz, Gilberto R. de (INPE),
%B Simpósio Brasileiro de Geoinformática, 18 (GEOINFO)
%C Salvador
%8 04-06 dez. 2017
%I Instituto Nacional de Pesquisas Espaciais (INPE)
%J São José dos Campos
%P 74-85
%S Anais
%X The most widely used change detection method is to classify remote sensing images independently for each date, and stack them to form a class sequence vector. However, impossible transitions within the sequences might occur and errors might be accumulated. To solve this, we propose a novel al- gorithm called Compound Maximum Likelihood (CML), based on the Maximum Likelihood classifier (ML). In CML information from all images is used jointly by considering the a priori probability of each class sequence. The algorithm was tested for Synthetic Aperture Radar and optical images classification in a study area in Para ́ state, within the Brazilian Amazon. CML presented either similar or very improved accuracy index values over ML land cover classifica- tions.
%@language pt
%3 8reis_escada.pdf