60016

Estimativa para identificação de áreas de pasto na Amazônia brasileira utilizando processamento de imagens de radar

Favoritar este trabalho

Abstract. The objective of this work is to show an automated method for determination of deforested areas based on images of Synthetic Aperture Radar (SAR) airborne, being used in subproject Land Cartography. This project aims to generate cartographic products of the area of almost 1.8 million square kilometers called "empty cartographic area” from the Legal Amazon region. This project is also known as "Radiography of the Amazon". Due to the large volume of data generated, a methodology for automatic identification of deforested areas is critical to the use of these inputs. Image Processing tools can contribute to the automatic extraction of information from SAR images, optimizing processing large amounts of data. The steps to the methodology for identifying areas of pasture in the Brazilian Amazon are by using low- pass filters, calculation of variance values of the digital values of pixels in the image and mathematical morphology. At the end, it is possible to verify that this methodology can identify deforested areas intended for farming, since these regions can be characterized by low values of the variance from the pixels’ values that represent it.