A Novel Approach for Change Detection Analysis of Land Cover from Multispectral FCC Optical Image using Machine Learning

Land covers refers to the physical land types such as vegetation, water, urban area, roads, and many more according to the geographical region. With the rapid change in land-use patterns, the land covers are varying drastically which requires immediate attention to have an eye at the impact of the land use planning and environmental changes is on the right track, or it needs to be modified. Hence utilizing the advancements in remote sensing technology for analyzing Land Use Land Cover (L ULC) classification maps using satellite images of the geographical region plays an important role in analyzing the present scenario of land covers. This paper proposes a novel approach for change detection analysis using the classification maps generated using Machine Learning (ML) classification techniques on a particular geographical region surrounding Nirma University, Ahmedabad, India. The highest classification accuracy of 98.48% was achieved using Support Vector Machine (SVM) for Near Infrared (NIR) band False Colour Composite (FCC) image obtained from Sentinel 2.