For scaling up low resolution multispectral images (LRMIs) with high resolution panchromatic image (HRPI), intensity-hue-saturation (IHS) can produce satisfactory spatial enhancement but usually introduces spectral distortion in the fused high resolution multispectral images (HRMIs). In this paper, to minimize this problem, we present a generalized intensity modulation (GIM) by extending the IHS transform to an arbitrary number of LRMIs, which uses the information of the spectral response functions (SRFs) of the multispectral and panchromatic sensors. Before modulation, the generalized intensity is enhanced by injecting details extracted from the HRPI by means of empirical mode decomposition. After the enhanced generalized intensity is substituted for the old one, the HRMIs are obtained through the GIM. Quickbird images are used to illustrate the superiority of this proposed method. Extensive comparison results based on visual analysis and Wald's protocol demonstrate that the proposed method is more encouraging for scaling up the LRMIs with the HRPI spectrally and spatially than the tested fusion methods.
Keywords: Empirical Mode Decomposition; Generalized Intensity Modulation; Intensity Hue Saturation; Scaling-up Transformation; Spectral Response Function.