Augmenting efficacy of polarization-based underwater image restoration through enhancement and color calibration

Underwater imaging is fundamental to underwater exploration and monitoring. However, due to the selective absorption by water and the scattering effects of suspended particles, imaging in turbid water often suffers from low contrast and color distortion. Particularly in extremely turbid underwater scenes, the degradation of image contrast and color fidelity is exacerbated, making the imaging quality inadequate for the requirements of scientific research and industrial production. This study builds upon previous findings in polarization-based underwater restoration. A novel approach termed block dilating histogram equalization with decremental constraints was employed to enhance the polarized underwater image pair. Color correction was refined through an optimized color restoration method that combines channel compensation with the shade of gray algorithm. Furthermore, in response to the typical presence of a small number of extreme pixel values in underwater images, a pixel tail-squeezing method was proposed to preprocess them. In addition, this study establishes the mathematical model for the three processing stages, which provides theoretical support and validation for the experimental outcomes. The experimental results demonstrate that pixel tail-squeezing can effectively reduce the influence of extreme values in the restoration process. Block dilating histogram equalization with decremental constraints enhanced low-contrast underwater images, managing to avoid noise amplification and preventing unreasonable exposure. The improved color correction method can more authentically restore the color of the enhanced original polarized images. Overall, this research combines theoretical analysis with experimental results to improve the restoration method based on the polarization restoration model. Through a series of preprocessing techniques, the key information of the original polarized image pair has been enhanced, leading to a significant improvement in the quality of the final restored images. Especially in extremely turbid water, the proposed method still works well. This study has guiding significance for the research on turbid underwater and foggy scene image restoration methods and practical application in scientific research and industrial fields.