Remote sensing image fusion is a technology to generate new image data using certain algorithms to combine the complementary information in image data obtained by the same/different remote sensor after these data's preprocess. It finds the missing spatial information in the multi-spectral images using the panchromatic ones and transfers this information into the multi-spectral images; finally producing the result images owning high spatial resolution and good spectral preservation. At present, many image fusion methods exist, such as IHS, PCA, Brovey transform and wavelet transform. Whereas the traditional IHS method only processes three bands, this paper addresses a generalized IHS (GIHS) image fusion method. It generalizes the traditional RGB to IHS transform from 3-D to N-D, in order to handle multiple bands. The proposed method is tested with ETM+ panchromatic and multi-spectral images and experimental results are evaluated visually and quantitatively. It shows that the GIHS method can gain sharp texture and more reservation of detail edge information compared to the traditional methods, and the spectral and radiometric distortions have no improvements, which needs more research.