Segmentação e classificação de imagens destinadas ao mapeamento digital do Parque Nacional da Serra dos Órgãos (PARNASO)

Abstract

In the environmental point, the National Park of Serra dos Órgãos (PARNASO) is the more studied National Park in Brazil and is configured as an important Unit Conservation in the Rio de Janeiro s state, because the environmental importance for the state. Located in four cities to the Serrana region: Teresópolis, Petrópolis, Magé and Guapimirim were evidenced that this area has passed for some problems, relatively recent, of disordered occupation because the urban expansion in its neighborhood characterized for human pressure. Through the digital image processing, more specifically the stages of segmentation and classification, were possible to illustrate the occupation process human being by means of cartographic documents. Beyond this process to make possible the generation of use the Land and vegetal covering maps, with the intention for assisting and giving foments to the execution of activities, the digital mapping is configured in an important tool for the environmental analysis, and the posterior zoning to the study area. Thematic classes of Land use and occupation with the intention had been adopted to allow the classification to the worked digital images. They are: rocky outcrop, urban area, agriculture and vegetation. Studies had been made in the direction to indicate and to explore the functionalities of tools SPRING and DEFINIENS and results they had been compared from use of images LANDSAT, CBERS, SPOT and IKONOS arriving itself resulted in SPRING system, the best parameters to be chosen they had been 10 similarity and area 400. Already for DEFINIENS system, it was evidenced that the segmentation process multilevel allowed the reach of faster results, to the computational view, to the only process of segmentation used normally enters the systems of digital image processing as the SPRING. Already to another view, the process of images classification, the research constituted in evaluating this mechanism by means of two pointers: accuracy/exactness and the Kappa index. In this direction, resulted trends of better had been observed in the SPRING system.