Quality Assessment of Pan-sharpening Methods

Kurzfassung

The quality of pan-sharpened image is usually quantified separately by various spectral and spatial quality measures mostly originating from image processing. This quantity and diversity of quality measures makes it quite difficult to rank different image fusion methods. A new Joint Quality Measure (JQM) is proposed which is based on the combination of two Structural SIMilarity (SSIM) indices (one for spectral quality and another one for spatial quality) allowing comparison of different methods using a sole measure. Quality assessment of four fusion methods: Component Substitution (CS), General Fusion Filtering (GFF), variant of GFF and ATWT (one of ARSIS implementations) is performed for IKONOS and WorldView-2 satellite optical remote sensing data. Experiments and results showed the superiority of the proposed JQM when compared with already known joint measure - Quality with No Reference (QNR) index. Moreover, the results are fully supported by a visual analysis of imagery.