摘 要：针对雾天拍摄图像不清晰的问题，提出了一种基于改进 AOD-Net 的单幅图像去雾算法。即在已有 AOD-Net 去雾算法的基础上，针对其存在的参数估计不均导致累计误差，从而使得去雾后的图像色彩失真以及低照度下效果不佳等缺陷，对AOD-Net 去雾算法的损失函数进行了改进，同时引入一个低照度增强模块。由实验结果可知，改进的方法在真实图像和人工合成图像上去雾效果都有了较大的提升，有效地提高了图像质量，在 PSNR 和 SSIM 指标上均优于其他对比算法。
关键词：图像去雾；AOD-Net；低照度增强模块；PSNR；SSIM
DOI:10.19850/j.cnki.2096-4706.2023.01.021
中图分类号：TP391 文献标识码：A 文章标号：2096-4706（2023）01-0080-04
Research on Improved Single Image Defogging Algorithm Based on AOD-Net
JIN Binfeng, HUANG Jinwei, ZHAO Huimeng
(School of Computer Science and Engineering, Anhui University of Science and Technology, Huainan 232001, China)
Abstract: In order to solve the problem of unclear images taken in fog, a single image defogging algorithm based on improved AOD-Net is proposed. That is, based on the existing AOD-Net defogging algorithm, the loss function of AOD-Net defogging algorithm is improved, and a low illumination enhancement module is introduced at the same time, aiming at the defects that the uneven parameter estimation leads to cumulative error, which leads to the color distortion of the image after defogging and the poor effect under low illumination. From the experimental results, we can see that the improved method has greatly improved the defogging effect in both real images and synthetic images, and effectively improved the image quality, and is superior to other comparison algorithms in terms of PSNR and SSIM indicators.
Keywords: image defogging; AOD-Net; Low illumination enhancement module; PSNR; SSIM
参考文献：
[1] RAHMAN Z，JOBSON D J，WOODELL G A. Multi-scale Ret-inex for color image enhancement [C]//Proceedings of 3rd IEEE International Conference on Image Processing.Lausanne：IEEE， 1996：1003-1006.
[2] THANH L T，THANH D N H，HUE N M，et al. Single image dehazing based on adaptive histogram equalization and linearization of gamma correction [C]//2019 25th Asia － Pacific Conference on Communications（APCC）．Ho Chi Minh City： IEEE，2019：36-40．
[3] HE K M，SUN J，TANG X O，et al. Single image haze removal using dark channel prior [EB/OL].[2022-08-29].https://www. docin.com/p-824571919.html.
[4] HE K M，SUN J，TANG X O. Guided image filtering [J]. IEEE transactions on pattern analysis and machine intelligence，2013， 35（6）：1397-1409.
[5] CAI B L，XU X M，JIA K，et al. DehazeNet：An End-to-End System for Single Image Haze Removal [J].IEEE Transactions on Image Processing，2016，25（11）：5187-5198.
[6] ZHANG J，TAO D C． FAMED － Net：AFast and Accurate Multi-Scale End － to-End Dehazing Network [J].IEEE Transactions on Image Processing，2020，29（1）：72-84.
[7] LI B Y，PENG X L，WANG Z Y，et al． AOD-net：Allin-one dehazing network [C]//2017 IEEE International Conference on Computer Vision（ICCV）.Vision：IEEE，2017：4770 － 4778.
[8] ZHU A Q，ZHANG L，SHEN Y，et al.Zero-Shot Restoration of Underexposed Images via Robust Retinex Decomposition [J].2020 IEEE International Conference on Multimedia and Expo（ICME）. London：IEEE，2020：1-6.
作者简介：金彬峰（1997—），男，汉族，浙江湖州人，硕士在读，研究方向：图像处理；黄金炜（1999—），男，汉族，江苏南通人，硕士在读，研究方向：图像处理；赵慧勐（2000—），男，汉族，山东济宁人，硕士在读，研究方向：图像处理。