This paper proposes a Fourier Transform-Based Dual-Branch Low-Light Image Enhancement Network, FTDB-Net, aimed at addressing issues such as color distortion, detail blurring, and noise in low-light conditions. By applying discrete Fourier transform, the image is divided into high-frequency and low-frequency regions. The high-frequency region contains rich details but is difficult to recover, relying on global information, while the low-frequency region is smoother and can be enhanced more easily using local information. FTDB-Net employs a dual-branch structure that combines the advantage of Transformers in capturing global information with the efficiency of convolutional networks in processing local information, optimizing the high-frequency and low-frequency regions separately. These two types of information are then combined through an adaptive fusion mechanism, significantly improving image clarity and detail. Experimental results show that FTDB-Net outperforms existing advanced methods on multiple benchmark datasets.