Images captured under low-light conditions often suffer from insufficient brightness, low contrast, limited gray-level range, color distortion, and severe noise. These degradations not only impair human visual perception but also significantly constrain the performance of various computer vision systems. To address the issue of limited color range in conventional HSV transformations for low-light image processing, we propose a low-light image enhancement network based on Decoupled HSV Representation Learning Network (DHRNet). Specifically, the proposed method employs a deep neural network to learn a nonlinear mapping from RGB to HSV color space, effectively overcoming the “dead zone” problem of traditional HSV conversion and enabling a more adaptive color-space decomposition tailored to low-light characteristics. On this basis, we develop an end-to-end enhancement network with feature mapping, decomposition, enhancement, and reconstruction modules. In addition, a hue–saturation polar coordinate loss is introduced to model human color perception and mitigate color shifts. Extensive experiments demonstrate that DHRNet achieves superior performance in low-light image enhancement, consistently outperforming state-of-the-art methods in qualitative evaluation, quantitative assessment, ablation studies, and generalization tests.