一种基于深度估计模型的图像加雾软件
An Image Fogging Software Based on Depth Estimation Model
王旭光 1张崇 1田珊珊 1白康1
作者信息
- 1. 华北电力大学河北省发电过程仿真与优化控制工程技术研究中心,河北 保定 071003
- 折叠
摘要
随着深度学习方面的研究不断深入,数据的重要性也上升到新的高度.在较为火热的图像去雾领域,去雾模型性能强弱很大程度上取决于雾天图像数据集的好坏.而现有数据集难以满足科研的需求,同时制作这样的数据集代价高昂,因此利用无雾图像合成真实雾天图像的研究显得尤为重要.在大气散射模型的基础上,引入单目图像深度估计的方法,实现无雾图像加雾的处理,同时比较了深度估计方法的鲁棒性,推荐了适合用于图像加雾的深度估计方法,实现了操作简单的图像加雾软件.
Abstract
With the deepening of deep learning research,the importance of data has risen to a new height.In the hot field of image defogging,the performance of the defogging model largely depends on the quality of the image data-set on foggy days.However,the existing data sets are difficult to meet the needs of scientific research,and the cost of making such data sets is high.Therefore,it is particularly important to synthesize real foggy images using fog-free im-ages.Based on the atmospheric scattering model,the method of monocular image depth estimation is introduced to re-alize the fogging process of fogless images.Meanwhile,the robustness of the depth estimation method is compared,and the depth estimation method suitable for image fogging is recommended.Finally,the image fogging software with sim-ple operation is realized.
关键词
大气散射模型/单目图像深度估计/图像合成雾/图像加雾软件Key words
Atmospheric scattering model/Monocular image depth estimation/Generate fog image/Image fogging software引用本文复制引用
出版年
2024
NETL
NSTL
维普
万方数据