基于双向多尺度特征融合的湍流退化图像快速复原  

Rapid Restoration of Turbulent Degraded Images Based on Bidirectional Multi-Scale Feature Fusion

作　　者：郭一鸣 吴晓庆[1,3] 苏昶东 张世泰 毕翠翠 陶志炜 Guo Yiming;Wu Xiaoqing;Su Changdong;Zhang Shitai;Bi Cuicui;Tao Zhiwei(Key Laboratory of Atmospheric Optics,Hefei Institutes of Physical Science,Chinese Academy of Sciences,Hefei 230031,Anhui,China;University of Science and Technology of China,Hefei 230026,Anhui,China;Advanced Laser Technology Laboratory of Anhui Province,Hefei 230037,Anhui,China)

机构地区：[1]中国科学院合肥物质科学研究院大气光学重点实验室,安徽合肥230031 [2]中国科学技术大学,安徽合肥230026 [3]先进激光技术安徽省实验室,安徽合肥230037

出　　处：《激光与光电子学进展》2022年第22期55-65,共11页Laser & Optoelectronics Progress

基　　金：国家自然科学基金(91752103)。

摘　　要：提出一种基于双向多尺度特征融合的生成对抗网络(GAN),利用该网络对各种地基望远镜拍摄的受大气湍流影响的目标天体图像直接进行盲复原处理。首先通过长曝光大气湍流退化模型与清晰图片进行卷积来构建数据集,并进行网络训练,在模拟湍流图像数据集中测试网络性能。同时,实际获取了Munin地基望远镜(卡塞格林型望远镜)拍摄的受湍流影响的国际空间站图片,并用所提神经网络模型进行测试。各项图像复原评价指标表明:所设计的网络实时性较强,在0.5 s内可以输出复原结果,相比传统非神经网络复原方法要快10倍以上;所提网络的峰值信噪比(PSNR)提高2 dB~3 dB,结构相似性(SSIM)提高9.3%左右,对受真实湍流影响的退化图像也有较好的复原效果。This study proposes a generative adversarial network(GAN)based on bidirectional multi-scale feature fusion to reconstruct target celestial images captured by various ground-based telescopes,which are influenced by atmospheric turbulence.This approach first constructs a dataset for network training by convolving a long-exposure atmospheric turbulence degradation model with clear images and then validates the network’s performance on a simulated turbulence image dataset.Furthermore,images of the International Space Station collected by the Munin ground-based telescope(Cassegrain-type telescope)that were influenced by atmospheric turbulence are included in this study.These images were sent to the proposed neural network model for testing.Different image restoration assessment shows that the proposed network has a good real-time performance and can produce restoration results within 0.5 s,which is more than 10 times faster than standard nonneural network restoration approaches;the peak signal to noise ratio(PSNR)is improved by 2 dB-3 dB,and structural similarity(SSIM)is enhanced by 9.3%.Simultaneously,the proposed network has a pretty good restoration impact on degraded images that are influenced by real turbulence.

关 键 词：双向多尺度特征融合 神经网络 大气湍流退化模型 盲复原 

分 类 号：O435.1[机械工程—光学工程] TP391.4[理学—光学]