Underwater Image Enhancement by Using a Colour-Correction and Dehazing Approach
Shubhankar Nalawade, Mohit Dua · Procedia Computer Science · 2025
Capturing clear images underwater is difficult because light gets absorbed and scattered, resulting in degraded image quality with reduced visibility and colour inaccuracy. This research paper proposes a comprehensive approach that combines colour correction and dehazing techniques to enhance such deteriorated underwater images. In the colour correction step, principles of colour constancy are utilized to address colour distortions caused by underwater conditions. This correction significantly improves colour accuracy and enhances the visual fidelity of the image by first converting the RGB image into CIELab colour space and then performing white balancing in the CIELab colour space. This step successfully eliminates colour casts and restores the natural colours of underwater scenes, leading to visually pleasing and accurate representations. In the dehazing step, our method utilizes a dark channel prior-based approach to remove the haze by estimating the overall atmospheric light and the transmission map. The dehazing step significantly reduces haze, enhancing the visibility of submerged objects and fine details. Our technique has been evaluated subjectively and objectively, and the results have demonstrated its efficacy in reducing haze, correcting colour cast, and preserving the natural appearance of the image. The enhanced results consistently achieve great results for RMSE, PNSR, SSIM, and UCIQE, surpassing several existing underwater enhancement methods.