GALEIR: Global Atmospheric Light Estimation based Underwater Image Restoration
Sheezan Fayaz, Shabir A. Parah, G. J. Qureshi · 2022 IEEE 7th International conference for Convergence in Technology (I2CT) · 2022
This paper presents a modified version of the underwater dark channel prior algorithm for subaquatic image restoration based on Global Atmospheric Light Estimation for Underwater Image Restoration called as GALEIR. The proposed method uses the concept of underwater dark channel prior and attempts to increase the accuracy in atmospheric light estimation. Our methodology reduces the chances of inaccurate estimation of global atmospheric light by applying statistical mode operation on 1% of the brightest pixels chosen. Besides, a fast guided filter has been employed instead of complex Soft Matting for the transmission map refinement. The above-mentioned changes incorporated in the underwater dark channel prior algorithm increases the visual quality of the underwater image and consumes less computational time compared to iterative Soft Matting based restoration algorithms. We proffer the analysis of GALEIR by which its limitations and applicability are revealed. GALEIR has been evaluated on Underwater Colour Image Quality Evaluator (UCIQE), Blind/Referenceless Image Spatial Quality Evaluator (BRISQUE), Underwater Image Quality Metrics (UIQM), and time complexity. The proposed algorithm yields higher UCIQE and lower BRISQUE scores of about 1.231 and 23.86, respectively, performing better than the state-of-the-art techniques.