Land Use Land Cover Classification Using Multi-spectral Satellite Imagery

Land cover change detection using multispectral images is a critical application in remote sensing. It involves monitoring and assessing alterations in the Earth’s surface’s physical and biological characteristics over time. Multispectral images acquired by satellite sensors, which capture data across various electromagnetic spectrum bands, enable the identification and tracking of diverse land cover types, including forests, crops, urban areas, water bodies, and barren land. The fundamental concept behind change detection is the comparison of two temporally separated satellite images to identify discrepancies between the two timestamps. To achieve this, two distinct approaches have been implemented. The first approach utilizes K-Means clustering, which categorizes pixels in the images into clusters based on their spectral attributes, generating color maps that highlight areas of change. In the second approach, Gaussian Mixture Models (GMM) are employed to model pixel value distributions, resulting in similar color maps that identify regions of land cover change. These algorithms play a crucial role in producing visual representations of where alterations in land cover have occurred, facilitating applications in environmental management, urban planning, and agriculture. In summary, land cover change detection through multispectral images and remote sensing techniques is essential for monitoring Earth’s surface changes over time. Two approaches, K-Means clustering and Gaussian Mixture Models, have been implemented to identify changes and generate colour maps that highlight areas of interest in land cover transformations.