Journal of Applied Optics, Volume. 43, Issue 5, 913(2022)
Super-resolution reconstruction of fiber optic coil image based on lightweight network
Figures & Tables(10)
Fig. 3. Iterative curves of different degradation models at 4× scale factors for test set
Fig. 5. Visual comparison of different methods for 4× scale factors
Table 1. PSNR/SSIM values for four models at 4× scale factors
View tableView in ArticleTable 1. PSNR/SSIM values for four models at 4× scale factors
Models A B C Full Parameters/K 97 155 155 155 PSNR/dB 37.39 38.92 38.87 38.97 SSIM 0.8971 0.9763 0.9758 0.9769
Table 2. Comparison of average results of PSNR, SSIM and IFC for some advanced networks
View tableView in ArticleTable 2. Comparison of average results of PSNR, SSIM and IFC for some advanced networks
Method 2× 3× 4× PSNR/dB SSIM IFC PSNR/dB SSIM IFC PSNR/dB SSIM IFC Bicubic 38.85 0.9072 6.080 35.09 0.8603 3.147 32.84 0.8132 1.987 ScSR[ 3 ]40.03 0.9217 6.104 36.25 0.8757 3.212 35.19 0.8521 2.214 SelfExSR[ 4 ]40.98 0.9433 6.671 37.81 0.8993 4.185 35.95 0.8575 2.626 SRCNN[ 6 ]41.55 0.9596 6.285 38.74 0.9290 3.720 36.45 0.8978 2.321 VDSR[ 7 ]42.88 0.9811 6.326 39.22 0.9507 3.841 37.51 0.9283 2.370 DRRN[ 12 ]43.96 0.9834 6.494 40.96 0.9750 3.911 38.36 0.9532 2.436 LapSRN[ 18 ]43.99 0.9857 6.516 41.20 0.9761 4.106 38.61 0.9663 2.587 DPSR[ 19 ]44.04 0.9846 6.477 41.23 0.9757 4.043 38.65 0.9654 2.562 Ours 44.08 0.9858 6.592 41.35 0.9793 4.174 38.97 0.9769 2.601
Table 3. Running time and parameters of different algorithms
View tableView in ArticleTable 3. Running time and parameters of different algorithms
Methods SRCNN VDSR DRRN LapSRN DPSR Ours Parameters/K 57 665 297 813 820 155 Time/s 0.028 0.151 6.357 0.042 2.163 0.018
Tools
Get Citation
Copy Citation Text
Qianchuang ZHANG, Chenxia GUO, Ruifeng YANG, Xiaole CHEN. Super-resolution reconstruction of fiber optic coil image based on lightweight network[J]. Journal of Applied Optics, 2022, 43(5): 913
Paper Information
Category: OE INFORMATION ACQUISITION AND PROCESSING
Received: Jan. 18, 2022
Accepted: --
Published Online: Oct. 12, 2022
The Author Email: GUO Chenxia (guochenxia@nuc.edu.cn)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20