In order to obtain high-quality remote sensing imagery, it is necessary to use image fusion algorithms that combine the spectral information of multispectral images with the spatial resolution of panchromatic images, resulting in fused images that offer both high spatial resolution and rich spectral details. The spatial-spectral fusion method based on feature extraction and reconstruction has become a prominent research area. However, challenges such as the loss of critical features and inadequate integration of spatial information still persist. To address these challenges, this paper proposes a remote sensing image fusion method based on multi-scale feature reconstruction (MSFR-CNN). The method integrates multi-scale features of the source image at each stage of the feature reconstruction process and incorporates a multi-scale fusion module to effectively combine these features. The effectiveness of the model is validated using highresolution multi-modal satellite data from both qualitative and quantitative evaluations. Experimental results demonstrate that the model achieves high spectral fidelity and effective spatial information integration.
Wu et al. (Fri,) studied this question.