METHODOLOGIES FOR MONITORING ATLANTIC FOREST DERIVED FROM HIGH RESOLUTION SATELLITE IMAGES

Abstract

This research aimed to develop a methodology for semi-automatic classification of multitemporal pattern of Atlantic forest cover. The main objective was to detect changes in coverage that occurred in forest fragments derived from IKONOS high-resolution images. To do this you have selected the northern portion of the massif of Pedra Branca in the city of Rio de Janeiro where images were acquired of the years 2009 and 2010. The images were visually rated and ortorretificas. Later using geographic knowledge generated in the visual classification is created a semantic model of classification that has been implemented in software InterImage. The results were promising suggesting that integration of the two methods allows greater speed and reliability in the classification process.