Urban Land Use Land Cover Classification based on GF-6 Satellite Imagery and Multi-feature Optimization

Urban land use/land cover (LULC) classification has long been a hotspot for remote sensing applications. With high spatio-temporal resolution and multispectral, the recently launched GF-6 satellite provides ideal open imagery LULC mapping. In this study, we utilized multitemporal images to generate six types of features, including spectral bands, texture built-up, waterbody, vegetation, red-edge indices. The minimum Redundancy Maximum Relevance (mRMR) algorithm was employed optimize feature selection. Subsequently, Random Forest (RF) Extreme Gradient Boosting (XGBT) were assessed using different selections. Besides, various configurations designed comparison. results indicate that mRMR-based RF method achieved highest overall accuracy 91.37%. temporal indices important features urban contributed mainly grassland cropland. These supplement existing methods assist in improving mapping areas with complex landscapes.