Multi-focus image fusion based on Transformer and weighted guided filtering

In multi-focus image fusion tasks, traditional methods have the problem of uneven processing between focusing and defocusing boundary regions. To address the above problem, this paper proposes a novel multi-focus image fusion based on a Transformer and weighted guided filtering. In this paper, the generative adversarial network is the backbone network. The generator performs an end-to-end multi-focus image fusion task. The Transformer is used to obtain global dependencies and low-frequency spatial details. Then it uses a cross-domain crossattention mechanism to help generate the double branches to achieve the effect of information interaction, and obtains redundant information and complementary information. An improved guided filtering algorithm is used to fuse CbCr channels after RGB color space conversion. The experimental results show that the proposed algorithm is superior to other algorithms in both subjective visual evaluation and objective evaluation, and the image fusion quality is further improved.