Acta Photonica Sinica, Volume. 54, Issue 1, 0110003(2025)
Spatial-spectral Collaborative Unrolling Network for Pansharpening
Figures & Tables(10)
Fig. 1. Schematic diagram of spatial-spectral collaborative unrolling network
Fig. 5. Visual comparisons on typical GF-2 data at full resolution
Table 1. Numerical evaluations on WV-3 dataset
View tableView in ArticleTable 1. Numerical evaluations on WV-3 dataset
Methods WV-3 PSNR↑ RMSE↓ SAM↓ SSIM↑ Q8↑ ERGAS↓ GSA07 22.516 4 19.086 2 6.336 9 0.634 3 0.574 2 7.826 7 SFIM00 21.415 4 21.665 5 6.571 8 0.541 5 0.452 5 8.855 3 Wavelet09 21.446 4 21.588 3 8.611 6 0.565 6 0.527 1 9.154 5 PanFormer22 30.477 2 7.632 7 3.850 3 0.936 8 0.931 6 3.183 0 CTINN22 31.856 4 6.512 1 3.781 5 0.951 8 0.946 0 2.742 1 LightNet22 32.001 8 6.404 0 3.661 2 0.952 5 0.947 2 2.685 3 SFIIN22 31.658 7 6.662 0 3.735 7 0.949 2 0.943 5 2.801 6 MDCUN22 31.297 8 6.944 6 3.787 3 0.942 9 0.936 3 2.929 5 LGTEUN23 32.218 8 6.245 9 3.466 4 0.954 5 0.949 4 2.628 6 ABFNet23 32.254 0 6.220 7 3.460 6 0.955 0 0.949 7 2.612 7 BiMPan23 32.285 0 6.198 5 3.454 9 0.956 5 0.951 8 2.582 1 SCUN 32.405 0 6.113 5 3.449 5 0.957 5 0.953 0 2.561 0
Table 2. Numerical evaluations on WV-2 dataset
View tableView in ArticleTable 2. Numerical evaluations on WV-2 dataset
Methods WV-2 PSNR↑ RMSE↓ SAM↓ SSIM↑ Q4↑ ERGAS↓ GSA07 33.597 5 5.329 2 3.283 0 0.889 9 0.568 1 2.540 2 SFIM00 32.633 4 5.954 8 3.420 5 0.872 8 0.515 9 3.191 9 Wavelet09 32.199 2 6.260 1 3.655 5 0.850 0 0.457 7 3.379 9 PanFormer22 41.358 1 2.180 9 1.380 8 0.973 1 0.823 6 1.061 7 CTINN22 41.201 5 2.220 6 1.409 2 0.973 5 0.814 9 1.088 0 LightNet22 41.558 9 2.131 1 1.357 9 0.973 9 0.822 0 1.038 2 SFIIN22 41.948 9 2.037 5 1.312 1 0.975 2 0.810 8 1.008 4 MDCUN22 42.335 1 1.948 9 1.237 6 0.977 2 0.837 0 0.963 8 LGTEUN23 42.624 0 1.885 1 1.208 9 0.978 2 0.840 1 0.928 2 ABFNet23 41.637 1 2.111 9 1.340 7 0.974 6 0.824 1 1.031 6 BiMPan23 42.272 9 1.962 8 1.243 3 0.977 1 0.838 0 0.964 9 SCUN 42.729 7 1.862 3 1.199 8 0.979 1 0.842 3 0.913 9
Table 3. Numerical evaluations on GF-2 dataset
View tableView in ArticleTable 3. Numerical evaluations on GF-2 dataset
Methods GF-2 PSNR↑ RMSE↓ SAM↓ SSIM↑ Q4↑ ERGAS↓ GSA07 36.055 7 4.015 6 3.672 7 0.883 8 0.551 7 3.575 8 SFIM00 34.771 5 4.655 4 3.764 3 0.857 2 0.458 0 4.207 3 Wavelet09 33.920 8 5.134 5 3.982 1 0.819 7 0.403 3 4.644 5 PanFormer22 44.854 0 1.458 3 1.552 7 0.980 5 0.886 5 1.333 4 CTINN22 44.294 2 1.555 4 1.678 8 0.978 4 0.871 6 1.414 8 LightNet22 44.687 6 1.486 5 1.598 6 0.978 7 0.874 1 1.351 0 SFIIN22 44.724 8 1.480 1 1.604 3 0.980 2 0.872 1 1.336 1 MDCUN22 45.567 7 1.343 2 1.443 9 0.982 5 0.891 5 1.224 9 LGTEUN23 45.836 4 1.302 3 1.415 2 0.984 0 0.897 3 1.182 4 ABFNet23 45.230 8 1.396 4 1.512 6 0.981 5 0.883 6 1.271 0 BiMPan23 45.539 8 1.347 6 1.432 4 0.982 2 0.892 8 1.235 7 SCUN 46.634 5 1.187 9 1.325 4 0.986 3 0.908 5 1.078 4
Table 4. Ablation results of replacing adaptive degradation matrix estimation
View tableView in ArticleTable 4. Ablation results of replacing adaptive degradation matrix estimation
OR 3×3 ORMC TR 3×3 TRMC PSNR Params（M） FLOPs（G） √ 46.534 6 0.187 6 1.59 √ 46.579 8 0.190 2 1.65 √ 46.606 8 0.199 6 1.72 √ 46.634 5 0.200 6 1.82
Table 5. Ablation results of approximate operator module
View tableView in ArticleTable 5. Ablation results of approximate operator module
SPATB SPETB SACC PSNR Params（M） FLOPs（G） √ 46.358 8 0.190 1 1.65 √ 46.287 6 0.178 9 1.52 √ √ 46.462 5 0.198 5 1.75 √ √ √ 46.634 5 0.200 6 1.82
Tools
Get Citation
Copy Citation Text
Jianwei ZHENG, Hongyi XIA, Honghui XU. Spatial-spectral Collaborative Unrolling Network for Pansharpening[J]. Acta Photonica Sinica, 2025, 54(1): 0110003
Paper Information
Category:
Received: Jul. 2, 2024
Accepted: Sep. 2, 2024
Published Online: Mar. 5, 2025
The Author Email: XU Honghui (xhh@zjut.edu.cn)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20