Comparison of Global- and Local-scale Pansharpening for Rapid Assessment of Humanitarian Emergencies

We compared global-scale pansharpening of full- or large-scene QuickBird satellite imagery with local-scale pansharpening of a subset of these scenes by using modified intensity-hue-saturation (mills), principal component analysis (PGA), multiplicative (MP), and high pass filter (tiff) methods. The spectral properties of all pansharpened images were evaluated by using quantitative indices such as erreur relative globale adimensionnelle de synthese (ERGAs), correlation coefficient (cc), relative difference to mean (RDm), and relative difference to standard deviation (RDs). This study discovered that local-scale pansharpening was generally lower and higher than the global-scale approach in terms of ERGAS and cc, respectively. In particular, local-scale HPF produced pansharpening results very close to the original multispectral image with less than 0.18 percent and 0.07 percent of RDM and RDS, respectively. Local-scale fusion results with PCA and MP were similar to those of the global-scale approach. Local-scale pansharpening that uses very high spatial resolution imagery could lead to the rapid assessment of the magnitude and severity of humanitarian situations by producing a color image of high spatial resolution with reduced processing time.