Figure 2
From: Dual-branch feature encoding framework for infrared images super-resolution reconstruction
The global feature dependency integration module. The input features are passed through three parallel GFDI blocks and the outputs of all GFDI blocks are fused to have globally relevant features. The internal structure of each GFDI block is similar in that it has a parallel branching structure. Notably, the scale of the convolutional blocks in the internal branching structure of each GFDI block is different. The purpose of setting up convolutional blocks of different sizes is to establish connections between features of different scales to form global correlations.
