Urban Land Cover Classification of Multispectral Satellite Images Using Artificial Neural Networks

With the scientific evolution and accessibility of multispectral remote sensing data, there is an ever-growing demand to classify satellite images accurately to back directorial bodies in decision-making such as urbanization, transportation, healthcare, disaster management, etc. In this paper, multispectral Sentinel-2 imagery of dense urban areas is used to classify the urban land cover into five different land cover classes, namely water bodies, barren land, vegetative, road network, and buildings using an artificial neural network (ANN). The performance of the proposed method is assessed on different evaluation metrics and compared with other well-known machine learning classifiers such as MLC. The proposed method has obtained an overall accuracy of 98.67 which is higher than the MLC method and in the case of urban component classification, the proposed method has performed better than its counterpart.