In this work two images from Landsat 7 and Cbers 2 were analyzed in order to identify small areas of semideciduous forest and to evaluate the classification accuracy made by three different analysts. The study was carried out in Lavras region, MG, using the SPRING GIS with the appropriate functions to jufil the digital classification and visual inspection. The comparisons between the classifications and accuracy assessment procedures employed the overall accuracy, the user's accuracy, the producer's accuracy and the Kappa coefficient. The results showed that the overall accuracy were higher than 90% and the Kappa coefficient ranged from 50% to 77% when the Landsat and Cbers images were compared by different analysts. The fragments vegetation maps made from digital classification of Cbers and Landsat satellites images presented high percentage of common areas and analysts made different maps but, those one produced from Cbers satellite images were better than the other classifications.
remote sensing; Kappa index; semideciduous forest
