Novel Approach to Random-Valued Impulse Noise Removal in Underwater Color Images with Enhanced Decision-based Median Filtering

Underwater imaging is critical in applications like marine exploration, underwater robotics and biological research, yet it is often compromised by impulse noise and unique challenges such as poor visibility and color distortions across Red, Green and Blue (RGB) channels. This paper introduces an Enhanced Decision-based Median Filter (EDBMF), specifically designed to address these issues by efficiently restoring noisy underwater images while preserving critical image details. The proposed method integrates an adaptive noise-detection mechanism to accurately identify corrupted pixels and an optimized median filtering approach to selectively restore these pixels without degrading uncorrupted areas. By maintaining edge structures, textures and color consistency, the EDBMF ensures high-quality restoration. Comprehensive experiments conducted on underwater image datasets show that the EDBMF delivers approximately 37% improvements in performance metrics such as Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index Measure (SSIM) and Root Mean Square Error (RMSE) compared to state-of-the-art methods. These results underscore the algorithm’s capability to enhance underwater image clarity and fidelity, addressing a critical need in underwater imaging technology.