<p>Single image super-resolution (SR) reconstruction aims to reconstruct a high-resolution (HR) image from a low-resolution (LR) input image, which is useful in many fields such as surveillance, fire rescue, and remote sensing. However, most available SR reconstruction methods are developed primarily for visible images, making it challenging to achieve satisfactory results when applied to infrared images. For example, the reconstructed SR infrared images exhibit problems such as a lack of crucial details, excessive smoothness, and poor perceptual quality. To solve the aforementioned issues, we propose a new framework for infrared image SR reconstruction tasks called Topology-Aware Reconstruction Network (TopoRN), which includes a novel Topological Feature-Assisted Enhanced (TFAE) module and an innovative Topology Invariant Auxiliary Loss (TIAL) function. Specially, TFAE uses persistent homology (PH) to extract information about topological features hidden in infrared images and injects the information into the generator network via a feature adapter, resulting in more detailed reconstructed SR infrared images. Furthermore, the TIAL function aids in capturing the detailed differences between SR and HR infrared images from a topological perspective, significantly improving the perceptual quality of the reconstructed SR infrared images. Comprehensive qualitative and quantitative experiments on two public benchmarks demonstrate the effectiveness and superiority of our method.</p>