Hyperspectral and Multispectral image Fusion based on diffusion model and MergeCNN

In recent years, diffusion models have attracted attention for Hyperspectral and Multi-spectral image fusion or Pansharping. However, a little noise often remains in the images generated by diffusion models, which affects the final quality. To mitigate this issue, we propose a diffusion model-based method for hyperspectral and multispectral image fusion. We introduce the MergeCNN module, which excels in local feature extraction and feature representation through deep convolution, and has a significant advantage in noise suppression.The MergeCNN module combined with the diffusion model further optimises the details of the image and improves the quality of the image. Experimental results on two public datasets show that DIFF-Merge exhibits advanced performance compared to existing commonly used methods.