Data Driven Multiband Image Fusion That Preserves Wavelength-Specific Image Features

In this paper, we propose a data-driven multiband image fusion approach called FusionUNet aimed at preserving image features that are specific to the wavelengths of the source images. Specifically, we extract low-level wavelength-specific image features from the source images and use them to guide the training process of the U-Net structure designed to combine source images of various wavelengths. We conducted extensive experiments on public datasets, and the results demonstrate that our algorithm outperforms the state-of-the-art approaches in both qualitative and quantitative aspects.