The primary objective of infrared and visible image fusion is to produce high-quality fused images that exhibit salient targets and rich textures. However, information in visible images is often submerged in darkness and accompanied by noise in night scenes, which results in weak texture details and poor visual perceptions in the fusion images. To solve this problem, we propose an end-to-end transformer-based low-light infrared and visible image fusion method, named TLLFusion. Specifically, to alleviate the low illumination of visible images, an illumination enhancement module (ILEM) is utilized to enhance the illumination information of visible images. Subsequently, a hybrid convolution-transformer module (HCTM) is carefully designed to extract local features of images and model texture and illumination distribution from local to global perspectives. Furthermore, a hybrid attention module (HAM) is utilized to capture important information from different images. We conduct extensive experiments on LLVIP and \(\textrm{M}^3\) FD datasets. The experimental results demonstrate that our method achieves state-of-the-art performance, outperforming existing infrared and visible image fusion methods.