Underwater image enhancement based on dark channel prior and unsupervised color correction

Underwater images commonly suffer from low contrast, color distortion (blue-green shift), and blurred details due to light absorption and particle scattering in water, which severely affects underwater imaging quality. Based on this, this paper proposes an underwater image enhancement method based on unsupervised color correction and dark channel prior. First, the dark channel prior theory is utilized to eliminate the scattering effects caused by underwater particles, achieving underwater dehazing. Then, distance estimation is employed to perform region segmentation on the processed images, enhancing local adaptability for color correction. Finally, combining Gray-World theory with Bradford chromatic adaptation transform, color correction is completed in an unsupervised manner. Experimental results demonstrate that the proposed algorithm outperforms classical methods such as CLAHE and UDCP in both subjective visual perception and objective evaluation metrics (UIQM, UCIQE, CCF). The method exhibits outstanding performance in color correction, detail preservation, and computational efficiency, and can adaptively handle different underwater environments, providing high-quality image inputs for tasks such as underwater robot vision and marine organism detection.