Abstract: Underwater image enhancement techniques are crucial for the development of marine resources. To address the difficulties faced by existing underwater image enhancement methods in dealing with the complex underwater environments of the real world, especially in effectively enhancing challenging samples, we propose a method based on dual-domain improvement in content and style. Firstly, a simple and effective framework is designed to decouple underwater images into content features and style features, followed by reconstruction. For the content domain, a dual-domain adaptive high-frequency adjustment block is designed to enhance detail and suppress noise in the content features of the original underwater images. For the style domain, a style library is established to enable style transformation by matching the optimal style during inference. Finally, the content consistency loss and the reconstruction loss are used to supervise the model in completing the reconstruction and enhancement of underwater images. Experimental results on the publicly available underwater image datasets UIEB, EUVP, and RUIE show that our method outperforms compared methods in both subjective visual effects and objective evaluation metrics (PSNR, SSIM, UIQM, UCIQE, and MUSIQ). Notably, it demonstrates significant advantages in enhancing challenging samples.