%0 Journal Article
%4 sid.inpe.br/mtc-m21b/2014/11.18.23.59.17
%2 sid.inpe.br/mtc-m21b/2014/11.18.23.59.18
%@doi 10.1590/S1982-21702014000300034
%@issn 1413-4853
%F scopus 2014-11 BoggionePeCaFoBoPe:2014:EvSiIm
%T Evaluation of simulated images of MUX camera from CBERS-4 satellite for environmental analysis / Avaliação de imagens simuladas da câmera MUX do satélite CBERS-4 aplicadas à análise ambiental
%D 2014
%A Boggione, Giovanni de Araujo,
%A Pereira, G.,
%A Cardozo, Francielle da Silva,
%A Fonseca, Leila Maria Garcia,
%@affiliation Instituto Federal de Educação, Ciência e Tecnologia de Goiás (IFG/GO)
%@affiliation Universidade Federal de São João del- Rei (UFSJ)
%@affiliation Instituto Nacional de Pesquisas Espaciais - INPE, Caixa Postal 515São José dos Campos, SP, Brazil
%@affiliation Instituto Nacional de Pesquisas Espaciais (INPE)
%B Boletim de Ciências Geodésicas
%V 20
%N 3
%P 590-609
%K accuracy assessment, correlation, land cover, Landsat thematic mapper, NDVI, performance assessment, satellite imagery, spatial resolution, vegetation mapping.
%X Simulation methods of orbital images are usually applied to evaluate the performance of a specific sensor. From the use of these techniques, is possible to analyze and estimate the behavior of predict sensor images, allowing an analysis of future applications. In this context, is necessary the assessment of satellite-based images and the possible applications derived by CBERS-4, which should be released at the end of 2014 and will have a policy of free distribution of the data. Thus, this study aims at evaluating the potential of the camera CBERS-4 MUX with 20 m spatial resolution for land cover mapping. For this, images MUX are simulated from RapidEye image using filtering techniques based on the imaging process. To evaluate the simulation results, an image of the camera Landsat-5 TM is processed to produce a land cover and NDVI maps and compare them to the maps generated by the simulated CBERS-4 MUX image. The experiments show that the results obtained by simulated image MUX were very similar to the ones obtained by TM-5. Overall, the classifications of land cover for the MUX and TM sensors exhibit good agreement, with an overall accuracy of 87% and Kappa of 0.72. Also, we noticed that NDVI values estimated by the MUX are 25% higher than the values estimated by the TM and have a correlation of 85% (significant at 0.05, Student's t test).
%@language pt
%3 1413-4853-bcg-20-03-0590boggione.pdf