<p>Captured images under water scenarios usually suffer from degradation problems (such as low contrast, color distortion, and ambiguous details), due to the absorption, scattering, and others occurring in the ocean. Naturally, the demand for acquiring high-quality and resolution underwater images from low-resolution coordinates is increasing. However, most of exsiting methods are difficult to reconstruct the fine details and enhancement-super-resolution performance on real underwater imaging. In this paper, we propose an effective method to enhance the image visual quality and improve the spatial resolution of underwater images via attention-guided and detail-preserving learning, termed UAGDL-SESR. Specifically, we design a Multi-scale Spatial Adaptive Feature Module (MSAF) that enhances the diversity of feature representations by incorporating the characteristics of multi-scale features into a single-stream architecture. Combined with a spatial attention mechanism, the most valuable features extracted from multiple scale are dynamically selected and integrated. Meanwhile, we introduce a Dual Attention Guidance Module (DAGM) to discriminately learn and emphasize the potential useful features from piexl and channel aspects to enhance the model performance. Furthermore, we propose a Detail Enhancement and Fusion Module (DEFM) that effectively enhances the detail and texture of images by utilizing transmission map information. Experimental results on multiple datasets demonstrate that the proposed UAGDL-SESR method achieves a superior performance on the quality enhancement and super-resolution task, and significantly outperforms existing approaches regarding both subjective and objective evaluation metrics.</p>