Comparative between MLC and Random Forest classification algorithms using Sentinel-2B image
DOI:
https://doi.org/10.29327/248949.21.21-2Keywords:
Remote sensing, Digital image processing, VegetationAbstract
Abstract: The elaboration of land-use and land-cover maps using classifying algorithms is a technique that allows the continuous monitoring of natural resources on large scales and provides valuable information. There are several classifiers can be used, each with its specific premises. This study aimed to compare the performance of two different classifiers: the Maximum Likelihood Classification (MLC) and the Random Forest. The classifications were carried in southern Minas Gerais, in an area with agricultural and urban sites, using image from the Sentinel-2B. The Random Forest obtained the best performance between the two classifiers, with a Kappa index of 0.77, although it presented issues to detect smaller water bodies. Therefore, it is an algorithm indicated for a more accurate mapping of areas with different characteristics.