Identificação de áreas irrigadas por pivô central no Estado de Goiás através do uso de técnicas de visão computacional

Abstract

Remote sensing is a very important tool for monitoring land and water use in irrigation practices. The increasing technological advances have enabled the advent of new resources that help this task, either through the availability of high-resolution multispectral satellite images, or through the advancement of the methods used to interpret these data. Computer Vision has proven to be an important ally in the analysis of aerial images, serving as a decision support tool and an alternative to manual visual inspection methods, which are often laborious. Based on these concepts, this work aimed to identify central pivot irrigated areas in the State of Goiás, using an automatic circular shape detection method. For this, we used the Hough Transform, an algorithm widely used in Computer Vision to detect geometric shapes, including circles. Applying the technique to satellite images of agricultural areas in the municipality of Cristalina – GO, the pivot circles could be automatically detected. The number of pivots counted manually was compared with the quantity detected automatically, in order to determine the detection accuracy of the algorithm.