High-resolution images have higher pixel density and more information. Super-resolution reconstruction based on image fusion uses two or more methods to fuse their images. That is to say, multiple images are synthesized into a new image by using the temporal and spatial correlation and complementarity of images. The composited image is fuller and clearer, with different details. The author draws on the idea of ensemble learning, and conducts experiments on the three algorithms of BasicaSR, SRGAN, and ESRGAN for pairwise cross fusion.
Image fusion is generally pixel-level fusion. Different methods assign different weights.
Summarize
1 Based on the advantages and disadvantages of different methods, ensemble learning is carried out.
2 Understand what image fusion is. Image fusion is to fuse images generated by different methods into one image.