水下光学图像质量评价体系综述

Comprehensive Review of Underwater Optical Image Quality Evaluation Systems

摘要:

旨在系统梳理水下光学图像质量评价方法的研究现状与发展趋势。方法上，从主观与客观评价 2 个角度综述现有水下图像质量评价（Underwater Image Quality Assessment，UIQA）方法，分析其原理、优缺点及适用性，并重点介绍了基于深度学习的新兴技术。结果表明，现有方法在应对水下图像的光散射、颜色失真与结构模糊等复杂退化问题上取得一定进展，但仍面临数据稀缺、模型泛化能力不足等挑战。未来需加强少样本学习、多模态融合与轻量化模型设计，以提升 UIQA 的准确性、鲁棒性与实时性，为水下图像处理与海洋观测应用提供有效支撑。

Abstract:

This paper aims to provide a systematic review of the current research status and development trends of underwater optical image quality assessment. Existing underwater image quality assessment(UIQA) approaches are reviewed from both subjective and objective perspectives. Their principles，advantages，limitations，and applicability are analyzed，with a focus on emerging deep learning-based techniques. The results indicate that while current methods have made certain progress in addressing complex degradation issues in underwater images such as light scattering，color distortion，and structural blur，they still face challenges including data scarcity and limited model generalization. In the future，it is necessary to enhance few-shot learning，multimodal fusion，and lightweight model design to improve the accuracy，robustness，and real-time performance of UIQA，thereby providing effective support for underwater image processing and marine observation applications.