Mapping vegetation, water bodies and urban areas in PeruSAT-1 satellite imagery

Abstract

This work proposes a classification model to identify (at a pixel level) relevant elements in PeruSAT-1 imagery, the first peruvian satellite acquired in 2016 by the Peruvian government. The relevant elements in remote sensing considered are vegetation, water, soil and urban areas. To perform the classification, a linear support vector machine (linear SVM) was trained using the information of 12'000 pixels and tested using 3'000. The proposed model achieves a 98.76% overall accuracy in the testing dataset.