The current panchromatic (PAN) / multispectral (MS) fusion methods do not comprehensively consider the characteristics of the remote sensing images from China’s domestic high-resolution satellites. Therefore, this paper proposes a variational fusion method for China’s domestic high-resolution images. On the one hand, the three-dimensional spectral high-fidelity model based on the spectral gradient is proposed by comprehensive consideration of the relations between the spectral bands. On the other hand, according to the existing blurring characteristics of the PAN image acquired by China’s domestic high-resolution satellites, the spatial enhancement model in consideration of the blurring degradation is developed. Finally, the fusion energy function is constructed by combining the prior knowledge of the remote sensing images, and it is solved by the classical gradient decent methods to obtain the fused image. The proposed method was tested and verified by the Gaofen-1 (GF-1) and Gaofen-2 (GF-2) satellite datasets. In addition, the popular GS, PRACS, and ATWT-M3 methods were applied for comparison from both qualitative and quantitative aspects. The experimental results show that the proposed variational high-fidelity PAN/MS fusion method comprehensively considers the characteristics of China’s domestic satellites, and hence it can maximally preserve the spectral information while effectively improve the spatial resolution of the MS images, thus achieving the best fused results.
Fig.1 Fusion results of the GF-1 simulated experiment
Fig.2 Fusion results of the GF-2 simulated experiment
评价指标
融合方法
GS方法
PRACS方法
ATWT-M3方法
基于传统光谱 保真项融合方法
本文方法
CC
0.841 7
0.857 9
0.860 7
0.884 8
0.894 0
PSNR
24.246 5
25.101 3
25.112 1
25.394 7
25.610 3
ERGAS
6.340 5
5.838 7
5.891 9
5.580 7
5.462 1
SAM
4.885 5
4.316 0
4.574 7
4.529 5
4.204 1
Tab.1 Quantitative evaluation of the fusion results in the GF-1 simulated experiment
评价指标
融合方法
GS方法
PRACS方法
ATWT-M3方法
基于传统光谱 保真项融合方法
本文方法
CC
0.979 4
0.976 6
0.973 3
0.990 1
0.990 8
PSNR
39.291 9
39.194 3
38.721 3
43.675 0
43.800 5
ERGAS
0.672 1
0.707 4
0.779 7
0.423 6
0.416 2
SAM
0.547 5
0.530 2
0.645 7
0.459 1
0.437 9
Tab.2 Quantitative evaluation of the fusion results in the GF-2 simulated experiment
Fig.3 Fusion results of the GF-1 real experiment
Fig.4 Fusion results of the GF-2 real experiment
评价指标
融合方法
GS方法
PRACS方法
ATWT-M3方法
基于传统光谱 保真项融合方法
本文方法
CC
0.798 0
0.996 4
0.994 2
0.997 4
0.998 0
PSNR
25.913 0
44.590 5
42.419 0
45.775 1
46.892 2
ERGAS
4.239 4
0.542 2
0.697 5
0.449 6
0.403 2
SAM
6.218 9
0.426 4
0.581 9
0.406 0
0.328 7
Tab.3 Quantitative evaluation of the fusion results in the GF-1 real experiment
评价指标
融合方法
GS方法
PRACS方法
ATWT-M3方法
基于传统光谱 保真项融合方法
本文方法
CC
0.948 4
0.988 1
0.991 4
0.992 0
0.994 0
PSNR
30.268 1
38.633 9
39.046 6
39.441 1
40.442 4
ERGAS
1.153 5
0.491 7
0.492 1
0.445 9
0.391 6
SAM
0.349 9
0.290 2
0.355 0
0.418 2
0.250 8
Tab.4 Quantitative evaluation of the fusion results in the GF-2 real experiment
[1]
Shen H F, Meng X C, Zhang L P . An integrated framework for the spatio-temporal-spectral fusion of remote sensing images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2016,54(12):7135-7148.
doi: 10.1109/TGRS.2016.2596290
url: http://ieeexplore.ieee.org/document/7560877/
Zhang L P, Shen H F . Progress and future of remote sensing data fusion[J]. Journal of Remote Sensing, 2016,20(5):1050-1061.
[3]
Aiazzi B, Alparone L, Baronti S , et al. Twenty-five years of pansharpening:A critical review and new developments[M] //Chen C H.Signal and Image Processing for Remote Sensing. 2nd edition.Boca Raton,FL: CRC Press, 2012: 533-548.
Chavez Jr P S, Sides S C, Anderson J A . Comparison of three different methods to merge multiresolution and multispectral data: Landsat TM and SPOT panchromatic[J]. Photogrammetric Engineering and Remote Sensing, 1991,57(3):295-303.
doi: 10.1163/157006707X222768
url: http://europepmc.org/abstract/AGR/IND91014290
[6]
Laben C A, Brower B V .Process for enhancing the spatial resolution of multispectral imagery using pan-sharpening:United States, 6011875[P].2000 -01-04.
Li W J, Wen W P, Wang Q H . A study of remote sensing image fusion method based on Contourlet transform[J]. Remote Sensing for Land and Resources, 2015,27(2):44-50.doi: 10.6046/gtzyyg.2015.02.07.
Meng X C, Shen H F, Li H F, et al. Improving the spatial resolution of hyperspectral image using panchromatic and multispectral images:An integrated method [C]//Proceedings of the 7th Workshop on Hyperspectral Image and Signal Processing:Evolution in Remote Sensing (WHISPERS).Tokyo,Japan:IEEE, 2015: 1-4.
Meng X C, Shen H F, Zhang H Y , et al. Maximum a posteriori fusion method based on gradient consistency constraint for multispectral/panchromatic remote sensing images[J]. Spectroscopy and Spectral Analysis, 2014,34(5):1332-1337.
[14]
Meng X C, Shen H F, Zhang L P, et al. A unified framework for spatio-temporal-spectral fusion of remote sensing images [C]//Proceedings of IEEE International Geoscience and Remote Sensing Symposium (IGARSS).Milan,Italy:IEEE, 2015: 2584-2587.
Wald L. Quality of high resolution synconfproced images: Is there a simple criterion? [C]//Proceedings of the 3rd Conference “Fusion of Earth Data:Merging Point Measurements,Raster Maps and Remotely Sensed Images”. Sophia Antipolis,France:SEE/URISCA, 2000: 99-103.