Efficient UAV-Borne Hyperspectral Image Classification Using Spatial Guided Filtering

This paper introduces a spectral and spatial classification approach specifically tailored for high spatial resolution UAV-borne hyperspectral remote sensing imagery. The proposed approach combines a support vector machine (SVM) classifier with a guided image filter to effectively utilize both spectral and spatial features. The guided filter is employed to extract spatial information and optimize classification results. Empirical results demonstrate a significant improvement in overall accuracy within a reasonable timeframe. These findings highlight the effectiveness of the proposed approach in enhancing classification accuracy, particularly in the context of Land Use and Land Cover applications (LULC). By leveraging the synergy between the SVM classifier and guided image filter, the method provides a practical solution for high spatial resolution hyperspectral image classification. Future investigations should focus on expanding the training dataset and assessing the classifier's performance on diverse hyperspectral images to enhance its adaptability and generalization capabilities.