基于深度学习的水面图像去雾算法综述

Review on Deep Learning-Based Water Surface Image Defogging Algorithms

摘要:

系统性的综述了基于深度学习的水面图像去雾算法。首先介绍去雾物理模型，明确雾天图像的基本特性；接着分类阐述当前基于深度学习的通用去雾算法，分析其在水面场景的适配情况；其次，对水域雾天的基本特性进行分析，梳理了常见公开的数据集与评价指标；然后总结水面去雾算法的最新研究进展；最后对该领域未来的研究方向与发展趋势进行展望。

Abstract:

Systematically reviewed deep learning-based water surface image defogging algorithms. Firstly, the physical defogging model is introduced, and the basic characteristics of foggy images are clarified. Secondly, the current deep learning-based general defogging algorithms are expounded by category, and their adaptation in water surface scenarios is analyzed. Thirdly, the basic characteristics of foggy conditions in water areas are analyzed, and common public datasets and evaluation indicators are sorted out. Then, the latest research progress of water surface defogging algorithms is summarized. Finally, the future research directions and development trends of this field are prospected.