Supervised and Unsupervised Classification Based on Remote Sensing for Study of an Area

Supervised and unsupervised classification, based on remote sensing, was used to study a territorial perimeter. NDWI, NDVI and NDBI indices were calculated based on satellite imagery. The NDVI variation relative to NDWI was described by a polynomial equation of degree 2, in statistical safety conditions (R-2 = 0.910, p <0.001). Unsupervised classification (U-class), based on the Iso Cluster algorithm (iterative process, which assigns each cell to a cluster based on Euclidean distances) and supervised classification (S-class), based on the Maximum likelihood algorithm (allocate each pixel to a class based on of maximum probability), led to the detection of three categories in the studied territory, with close values; water, vegetation, constructions (p <0.001 (95%). Remote sensing provides useful spectral information for the analysis of an area, and the operator can decide on one method or another of classification in relation to the additional information held in the territory under consideration.