<p>Images taken in low-light environments have poor visibility, low clarity, and obvious noise due to insufficient illumination, making the task of low-light image enhancement extremely challenging. Existing multimodal image enhancement methods are insufficient to restore areas of the image that lack effective information. Visible light and infrared image fusion methods can reveal hidden information in the dark through infrared images, but most existing methods only use it as auxiliary information and ignore intramodal enhancement, which limits the perceptual quality of the output image. Therefore, there are still challenges in balancing global and local illumination enhancement. To this end, we mainly introduce a new fusion network of visible and infrared images for low-light image enhancement. In order to reduce the impact of modal differences on cross-modal feature fusion, we propose a hierarchical cross-scale fusion (HCSF) module. This module takes full account of preserving global information in the fusion process. It also better utilizes the complementary information of the infrared image and the low-light image. At the same time, it greatly reduces the amount of computation. In addition, we propose a state-space feature module (SSFM). This module is used to fuse enhanced images and infrared weights in the feature reconstruction process. With this module, texture information can be enhanced at the pixel level. Extensive experiments on public datasets demonstrate that our method can outperform other state-of-the-art methods for low-light images.</p>