FusionImage: an R package for pan-sharpening images in open source software

Abstract

the objective of this article is to evaluate the performance of three pan-sharpening algorithms (high-pass filter, principal component analysis and Gram–Schmidt) to increase the spatial resolution of five types of multispectral images and to evaluate the results in terms of color, coherence and spatial sharpness, both qualitatively and quantitatively. A secondary objective is to present an implementation of the aforementioned pan-sharpening techniques within the open source software R. From a qualitative point of view, pan-sharpening of images with a high spatial resolution ratio give better results than those whose spatial resolution ratio is 2. According to the quantitative evaluation, there is no pan-sharpening methodology that obtains optimal results simultaneously for all types of images used. The results of the spectral and spatial ERGAS index vary for four out of the five types of images analyzed. The results show that none of the methods implemented in this work can be considered a priori better than the others. At the same time, this work indicates the importance of both qualitative and quantitative assessment.