12-19 December Full - page 641

larger spectral distortion than our method. The prediction
results of the proposed methods are more similar to the real
images in spectral information and spatial structure. The fusion
results are quantitatively evaluated using six evaluation indices
including RMSE , SSIM, r, UIQI, ERGAS, and SAM. The results are
shown in Table 1. It can be seen from the Table 1, although the
RMSE values of three bands in the fusion results of Fit-FC have
smaller RMSE than the proposed method, the RMSE values of
other bands show more obvious improvement for the results of
our method. Moreover, from other evaluation indices, the ER-
GAS and SAM of the proposed algorithm are the smallest among
the three algorithms, which shows that the fusion results of the
proposed algorithm are more similar to the real image. In ad-
dition to the spatial and spectral characteristics, the proposed
algorithm has higher SSIM, R, and UIQI compared with the other
two methods. It also shows that the fusion results of the pro-
posed algorithm have stronger correlation with real images, and
have better ability to capture the spatial structure of objects.
Experiment 2
In this experiment, the 1st , the 2nd, the 5th, the 6th, the 9th, the
10th, the 13th, and the 14 th images in the LGC data sets are set
as the training data. These images are synthesized into four
Figure 4. Observed: (a) Landsat image in 22 Feb 2002, (b) MODIS image in 22 Feb 2002 and predicted, (c) Landsat image in 10
Mar 2002, (d) MODIS image in 10 Mar 2002.
Figure 5. Predication results on 11 March 2002 fused by (a) actual image, (b) STARFM , (c)Fit-FC, and (d) our method" to " (a)
actual image on 11 March 2002, and predication results fused by (b) STARFM, (c) Fit-FC, and (d) our method.
Table 1. Accuracy evaluation of fusion results of STARFM,
Fit-FC, and the proposed algorithm in Experiment 1.
Band RMSE SSIM r QI ERGAS SAM
STARFM
1 0.0116 0.9518 0.8356 0.4361
0.8766 4.6106
2 0.0152 0.9353 0.8122 0.4635
3 0.0257 0.8784 0.8348 0.4704
4 0.0371 0.8012 0.8697 0.4846
5 0.0405 0.7686 0.8964 0.4581
7 0.0431 0.7522 0.8764 0.4523
Fit-FC
1 0.0096 0.9564 0.8822 0.3446
0.7469 4.6209
2 0.0126 0.9445 0.8612 0.3870
3 0.0200 0.8959 0.8912 0.4624
4 0.0361 0.8452 0.8761 0.5164
5 0.0389 0.7968 0.9042 0.4344
7 0.0368 0.7868 0.9048 0.4426
Ours
1 0.0102 0.9566 0.8750 0.3935
0.7377 3.8808
2 0.0134 0.9433 0.8534 0.4625
3 0.0205 0.9028 0.8983 0.5225
4 0.0322 0.8561 0.9016 0.5476
5 0.0364 0.8253 0.9126 0.5183
7 0.0340 0.8273 0.9209 0.5323
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING December 2019 911