Satellite Image Analysis for Land Use and Change Detection

Abstract

This paper presents a comprehensive approach to satellite image analysis for land use classification and environmental change detection. By leveraging multispectral imagery from Sentinel-2 and Landsat 8/9, the proposed model automates the classification of diverse geographic regions, such as water bodies, forests, urban areas, and agricultural lands. The methodology incorporates advanced deep learning architectures, specifically Convolutional Neural Networks (CNNs) and U-Net, for robust feature extraction and pixel-level semantic segmentation. Furthermore, a change detection framework is introduced to identify significant environmental transformations over time-series data, highlighting critical issues like deforestation and urban expansion. The integration of geospatial data handling libraries ensures efficient preprocessing, while the outcome provides actionable insights visualized through comprehensive map overlays. This research demonstrates the efficacy of deep learning in remote sensing and its potential for real-time environmental monitoring.