%0 Journal Article
%T 基于深度学习的图像去雾研究综述<br>A Review of Research on Image Defogging Based on Deep Learning
%A 吉波涛
%A 陆利坤
%J Journal of Image and Signal Processing
%P 21-33
%@ 2325-6745
%D 2025
%I Hans Publishing
%R 10.12677/jisp.2025.141003
%X 恶劣天气环境下拍摄的图像会受到雾或霾的影响&#65292;从而导致图像饱和度过低模糊、以及颜色灰白等负面效果&#65292;这不仅会使图像中的重要信息丢失&#65292;还会对后续计算机视觉任务(如目标检测、图像分割、人员再识别)的研究造成负面影响。为了解决上述问题&#65292;文章首先对图像去雾的发展历程进行分析和梳理&#65292;接下来重点论述了深度学习在图像去雾领域的研究进展&#65292;主要包含有监督去雾、无监督去雾和半监督去雾技术&#65292;并对各自的代表性算法进行深入对比分析。最后&#65292;介绍了图像去雾领域主流的数据集和评估指标。<br>Images captured in harsh weather environments are often affected by fog or haze, which can lead to negative effects such as low saturation, blurring, and grayish-white colors. This not only results in the loss of important information in the image, but also has a negative impact on subsequent computer vision tasks such as object detection, image segmentation, and personnel re-identification. This article first provides a comprehensive analysis and sorting of image defogging and then reviews the research progress of deep learning in the field of image defogging, mainly including supervised defogging, unsupervised defogging, and semi-supervised defogging. We compared and analyzed representative algorithms among these methods. Finally, the commonly used datasets and evaluation metrics for image defogging were introduced.
%K 深度学习&#65292
%K 单幅图像去雾&#65292
%K 图像处理<br>Deep Learning
%K Single Image Defogging
%K Image Processing
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=104598