Avaliação de Tipos de Classificações em Imagens de Média e Alta Resolução Espacial, no Bioma Cerrado

Abstract

The Cerrado is a biome of great importance with a high degree of biodiversity, however, it is in degradation process. Established tools for evaluating this process are the remote sensing together with geotechnologies. Thus is possible to generate maps of land use whith constant monitoring of the modifying processes of land cover. In this article, remote sensing techniques are used to classify, through three different methods, an image of the Landsat sensor for the municipality of Dois Irmaos do Buriti, located in Mato Grosso do Sul State. It was verified that the unsupervised method presented problems in the final process, making their employment unfeasible. On the other hand the Object-Oriented Image Classification and Supervised Classification Methods shows moderate results with better practiced in the Supervised Classification. Another objective was to evaluate the adaptation of the key for classifying. Landsat in the areas of Cerrado with the an image of Alos sensor that has high spatial resolution, but lower spectral resolution. The result corroborates for an applicability of the classification keys in the images Alos, as long as they are considered the respective areas, especially do not refer to the absence of short wave infrared in the AVNIR-2 sensor, which can be seen in a deficient delimitation of agricultural areas.