Haze removal is an interesting topic in multimedia and image processing for many applications. Specially for the automatic piloting of ships, the haze removal technology for aqueous vapour regions plays a key role in safe piloting. However, the existing haze removal methods did not dehaze well for these areas. Based on this motive, this paper presents a new haze removal approach to improve the dehazing effect for aqueous vapour regions, in which we design two new computing mechanisms. The first one is to propose a new gradient change model of the dark channel value related to aqueous vapour regions. The second one is to design an optimized and iterated correction method for the dark channel of aqueous vapour regions. Finally, based on these two computing mechanisms, a dynamic iterative optimal correction model is presented to solve the proposed method. Both the visual and the quantitative experiments demonstrate the proposed method outperforms both the family methods of dark channel prior and the deep learning-based methods in aqueous vapour regions. In conclusion, the proposed method can effectively remove the haze in aqueous vapour regions.

去雾是多媒体和图像处理中许多应用的一个有趣话题。尤其是船舶自动引航，水汽区除雾技术对安全引航起到关键作用。然而，现有的除雾方法对这些区域的除雾效果不佳。基于这个动机，本文提出了一种新的去雾方法来改善水蒸气区域的去雾效果，其中我们设计了两种新的计算机制。第一个是提出与水蒸气区域相关的暗通道值的新梯度变化模型。二是针对水汽区暗通道设计优化迭代校正方法。最后，基于这两种计算机制，提出了一种动态迭代最优修正模型来求解所提出的方法。视觉和定量实验都表明，所提出的方法优于暗通道先验家族方法和基于深度学习的水蒸气区域方法。总之，所提出的方法可以有效去除水蒸气区域的雾度。