Advanced Imaging, Volume. 1, Issue 2, 021003(2024)
Ultra-robust imaging restoration of intrinsic deterioration in graded-index imaging systems enabled by classified-cascaded convolutional neural networks
Figures & Tables(7)
Fig. 2. Image restoration by (a) the single CNN and (b) the CC-CNN.
Fig. 3. Details of the implemented networks. (a) ResNeSt50 image classification. (b) U-Net image restoration.
Fig. 4. Comparison of restoration results between the single CNN and the CC-CNN for virtual imaging.
Fig. 5. Imaging restoration for each GRIN lens by the single CNN and the CC-CNN.
Table 1. Comparisons between Single U-net and CC-CNN in All Images.
View tableView in ArticleTable 1. Comparisons between Single U-net and CC-CNN in All Images.
Network Type PSNR SSIM Random length U-Net 27.74 0.76 GRIN CC-CNN 30.97 0.81
Table 2. Comparisons between Single U-Net and CC-CNN in the Virtual Images.
View tableView in ArticleTable 2. Comparisons between Single U-Net and CC-CNN in the Virtual Images.
Network Type PSNR SSIM Random length U-Net 22.98 0.70 GRIN CC-CNN 31.00 0.81
Tools
Get Citation
Copy Citation Text
Zaipeng Duan, Yang Yang, Ruiqi Zhou, Jie Ma, Jiong Xiao, Zihang Liu, Feifei Hao, Jinwei Zeng, Jian Wang, "Ultra-robust imaging restoration of intrinsic deterioration in graded-index imaging systems enabled by classified-cascaded convolutional neural networks," Adv. Imaging 1, 021003 (2024)
Paper Information
Category: Research Article
Received: Jun. 15, 2024
Accepted: Aug. 19, 2024
Published Online: Sep. 20, 2024
The Author Email: Jinwei Zeng (zengjinwei@hust.edu.cn), Jian Wang (jwang@hust.edu.cn)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20