Underwater image enhancement via adaptive color-cast elimination and multi-weight pyramid fusion

To address the prevalent issues of color deviation, low contrast, and detail loss in underwater images, this study presents an integrated enhancement framework, termed ACE-Fusion, which combines adaptive color-cast elimination with multi-weight pyramid fusion. While building upon established techniques—such as color correction, CLAHE, gamma correction, guided filtering, and multi-scale fusion—this work introduces key innovations in their integration and parameterization. Specifically, an adaptive color-cast correction module is developed based on dual-threshold detection and iterative optimization, enhanced by a dynamic attenuation mechanism to prevent over-correction and artifacts. In the Lab color space, the luminance channel of the original image is decomposed and refined through guided filtering to enhance high-frequency details. Global contrast is further improved by integrating CLAHE with adaptive gamma correction, ensuring both texture preservation and brightness consistency. Finally, a multi-weight pyramid fusion strategy combines complementary information from detail-enhanced and contrast-improved images, achieving a balanced enhancement in color fidelity, detail clarity, and illumination uniformity. Experimental results on UIEB, EUVP, and UFO datasets, as well as real-world field tests on underwater vehicles, demonstrate that the proposed framework, ACE_Fusion significantly outperforms state-of-the-art algorithms in terms of PIQE and computational efficiency. Furthermore, demonstrates substantial application potential in marine engineering applications, particularly for underwater vehicles requiring real-time visual perception capabilities.