1 图像建模
1.1 分焦平面图像的场景信息提取
1.2 图像建模
2 重构算法
2.1 RLR-MDF高分辨率重构算法
2.2 LS-MDF高分辨率重构算法
3 实验结果
3.1 多光谱图像重构结果
图6 25个波段下基于LS-MDF方法的重构图像与参考图像的差分图注:网络版为彩图。 Fig. 6 Differential images between the reconstructed images based on LS-MDF method and the reference images of 25 wavebands |
图8 场景的待重构图像(a)、基于CNMF方法获得的重构图像(b)、基于Hysure方法获得的重构图像(c)、基于LS-MDF方法获得的重构图像(d)、基于RLR-MDF方法获得的重构图像(e)和参考图像(f) Fig.8 Image to be reconstructed(a), reconstructed image based on CNMF method(b), reconstructed image based on Hysure method(c), reconstructed image based on LS-MDF method(d), reconstructed image based on RLR-MDF method(e) and reference image(f) of scene |
表1 4种重构方法下的25波段重构图像的评价参数平均值Tab.1 Evaluation parameters’ mean values for 25-waveband reconstructed images using four reconstruction methods |
| 重构方法 | RMSE | ERGAS | CC | UIQI |
|---|---|---|---|---|
| CNMF | 0.135 08 | 5.631 5 | 0.627 14 | 0.589 23 |
| Hysure | 0.193 72 | 6.792 8 | 0.524 01 | 0.498 53 |
| LS-MDF | 0.126 05 | 5.510 3 | 0.684 45 | 0.652 47 |
| RLR-MDF | 0.118 73 | 5.194 6 | 0.732 79 | 0.642 57 |