Acta Photonica Sinica, Volume. 49, Issue 6, 0610001(2020)
Super-resolution in Digital Holographic Phase Cell Image Based on USENet
Figures & Tables(14)
Fig. 2. The general view of convolution layers for features extraction
Fig. 6. Fitting curves plotted by assessment scores from the input and output of 300 samples in validation set
Fig. 7. Fitting curves plotted by assessment scores from the input and output of 300 samples in validation set
Fig. 10. Fitting curves of SSIM performance in the ROI by the rebuilding of validation set
Table 1. Approaches with 15 different combinations of weighted loss factors
View tableView in ArticleTable 1. Approaches with 15 different combinations of weighted loss factors
ID α Proportion of L 1β Proportion of MSE γ Proportionof L 21 1 1.00 0 0.00 0 0.00 2 0.1 0.90 1 0.10 0 0.00 3 0.03 0.70 1 0.30 0 0.00 4 0.025 0.65 1 0.35 0 0.00 5 0.02 0.60 1 0.40 0 0.00 6 0.015 0.55 1 0.45 0 0.00 7 0.012 0.50 1 0.50 0 0.00 8 0.01 0.45 1 0.55 0 0.00 9 0.008 0.40 1 0.60 0 0.00 10 0.005 0.35 1 0.65 0 0.00 11 0.003 0.20 1 0.80 0 0.00 12 0 0.00 1 1.00 0 0.00 13 0 0.00 0 0.00 1 1.00 14 0.025 0.65 0.9 0.30 0.002 0.05 15 0.020 0.55 1 0.40 0.002 0.05
Table 2. The performances of net in convergence interval on validation set with 15 different loss functions
View tableView in ArticleTable 2. The performances of net in convergence interval on validation set with 15 different loss functions
ID SSIM_OPT SSIM_AVG MSE_OPT×10-3 MSE_AVG×10-3 PSNR_OPT PSNR_AVG 1 0.940 1 0.937 7 0.989 1.018 42.628 42.516 2 0.936 5 0.930 1 1.027 1.131 42.307 41.992 3 0.939 5 0.935 0 0.908 1.035 43.061 42.461 4 0.942 7 0.9398 0.847 0.862 43.403 43.308 5 0.941 1 0.9369 0.860 0.891 43.329 43.151 6 0.941 6 0.937 1 0.845 0.858 43.410 43.329 7 0.939 5 0.937 7 0.841 0.858 43.436 43.332 8 0.940 3 0.935 4 0.864 0.882 43.293 43.175 9 0.939 7 0.936 6 0.837 0.846 43.457 43.406 10 0.937 0 0.935 1 0.928 0.947 42.951 42.172 11 0.933 7 0.927 8 1.053 1.224 41.918 41.327 12 0.928 2 0.919 2 0.912 0.990 42.948 42.689 13 0.928 5 0.920 1 0.994 1.044 42.634 42.007 14 0.940 9 0.931 8 0.894 1.152 43.117 42.131 15 0.937 6 0.928 6 0.913 1.154 43.027 42.171
Table 3. The influence of ROI rebuilding by the calibration layer
View tableView in ArticleTable 3. The influence of ROI rebuilding by the calibration layer
Model Global (SSIM) ROI(SSIM) USENet 0.942 7 0.970 3 Reference 0.940 9 0.965 5
Tools
Get Citation
Copy Citation Text
Wen XIAO, Jie LI, Feng PAN, Shuang ZHAO. Super-resolution in Digital Holographic Phase Cell Image Based on USENet[J]. Acta Photonica Sinica, 2020, 49(6): 0610001
Paper Information
Category: Image Processing
Received: Jan. 19, 2020
Accepted: Apr. 16, 2020
Published Online: Nov. 26, 2020
The Author Email: PAN Feng (panfeng@buaa.edu.cn)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20