An Evaluation of Pan-Sharpening Methods forSuperView-1Satellite Imagery

Abstract

At present, little research focuses on the application of pan-sharpening methods to SuperView-1 satellite imagery. There is a lack of suitability assessment for existing pan-sharpening methods applied to SuperView-1 images. This study proposes an evaluation method that integrates visual evaluation, spectral analysis of typical objects, and quantitative indicators to evaluate the advantages of different pan-sharpening methods in different scenes of SuperView-1 imagery. Four scenes (urban areas, farmland, sparse vegetation, mixed surfaces) are selected to evaluate eight typical pan-sharpening methods (Brovey, principal component analysis (PCA), Gram-Schmidt (GS), band-dependent spatial-detail (BDSD), high-pass filtering (HPF), smoothing filter-based intensity modulation (SFIM), modulation transfer function-generalized Laplacian pyramid (MTF-GLP), MTF-GLP-high pass modulation (MTF-GLP-HPM). The results show that the suitability of each pan-sharpening method is different in various scenes. PCA, Brovey, and GS distort the spectral information greatly, and the stability of the pan-sharpening results in different scenes which are poor. BDSD has strong stability and can better balance the relationship between spectral distortion and spatial distortion in different scenes. The multi-resolution analysis method has better applicability and stability for SuperView-1 imagery, among which MTF-GLP and MTF-GLP-HPM perform better in the pan-sharpening results. This study provides a reference for the selection of pan-sharpening methods for SuperView-1 imagery in different application fields.