The role of image interpolation in pansharpening

Abstract

Pansharpening is an efficient way of producing images of higher spectral and spatial fidelity. Since pansharpening aims to generate a spectrally enhanced image with the same spatial detail content of the source panchromatic (PAN) image, the source multispectral (MS) image is upsampled to the size of the source PAN image prior to pansharpening. Several image interpolation algorithms have been proposed for this purpose, which may lead the analysts to a confusion as to which of these algorithms should be used for the best pansharpening performance. Hence, this study aimed to investigate the role of widely used image interpolation algorithms in the quality of the pansharpened images. For this purpose, the nearest neighbor interpolation, bilinear interpolation, bicubic interpolation, interpolation with a polynomial kernel of 23 coefficients (INTERP23) and cubic spline interpolation algorithms were tested through several pansharpening techniques on three test sites with different characteristics. Investigations revealed that upsampling the source MS images with the INTERP23 algorithm resulted in the pansharpened images with the optimum spectral and spatial quality.