Super-resolution of infrared images based on hierarchical distillation network

Aiming at the problem of low image resolution that easily occurs in the process of infrared images acquisition, this paper proposes a novel hierarchical distillation network to achieve infrared images super-resolution. By designing a cascaded residual distillation module, the negative impact of the over-deep network model is reduced; meanwhile, a dual-path feature fusion module is constructed to further enhance the feature expression capability of the network model. Experiments were conducted on public datasets and evaluated using two evaluation metrics, Peak Signal-to-Noise Ratio (PSNR) and Structure Similarity Index Measure (SSIM). The experimental results show that the method in this paper improves 1.97 and 0.033 in PSNR and SSIM, respectively, compared with RCAN, and generates images with high definition, strong structure and rich detail information.