Multi-Source Fusion Algorithms for Satellite Image Resolution Enhancement in Urban Land Cover Mapping of Ho Chi Minh City, Vietnam

This study explores pan-sharpening techniques to enhance the spatial resolution of Landsat 8-9 optical imagery and dual-polarized (VV/VH) SAR data for improved urban land cover classification. Mediumto low-resolution satellite images often pose challenges in accurately delineating specific urban features. Sentinel-1 SAR data provides critical surface parameters such as moisture and conductivity through backscatter analysis, while optical imagery captures spectral reflectance essential for land cover discrimination. By integrating SAR and optical datasets, the fused imagery benefits from both spectral richness and enhanced spatial detail. This research evaluates the performance of Gram-Schmidt (GS) and Principal Component Analysis (PCA) fusion methods in combining Sentinel-1 SAR and Landsat 8-9 imagery for urban land cover mapping in Ho Chi Minh City (2024). The Structural Similarity Index (SSIM) and bias analysis confirm that the fusion process effectively retains spectral integrity while enhancing spatial resolution to 10m, thereby improving the identification of surface features in urban environments.