Synthetic Sensor of Landsat 7 ETM+ Imagery to Compare and Evaluate Pan-sharpening Methods

Several studies and applications are present in literature to compare pan-sharpening methods: the greater part of them is based on indexes that relate radiometric characteristics of derived images to the correspondent original ones. Among others correlation index, RMSE (Root Mean Square Error), UIQI (Universal Image Quality Index) and RASE (Relative Average Spectral Error) are used, but evaluation is limited if there aren’t other multispectral images with higher geometric resolution to which compare the pan-sharpened results. In this paper Landsat 7 ETM+ artificial sensor was introduced to permit more reliable comparison and evaluation of pan-sharpening methods. Data concerning Campania region (Italy) were used: firstly Landsat 7 ETM+ panchromatic image and multispectral ones with acquisition bands within that of pan were processed to conduct their digital numbers to radiance values; then synthetic sensor was considered and geometric resolutions degraded by reducing original pixel dimensions 15 m × 15 m and 30 m × 30 m respectively to 30 m × 30 m and 60 m × 60 m. In other terms, new images were derived as equivalent products to those that should have been obtained by sensor with lower geometric resolutions but same acquisition bands of Landsat 7 ETM+ (panchromatic and multispectral). Pan-sharpening methods were applied to derived images and products compared with original data so to obtain a credible evaluation of performances of the implemented approaches.