Pansharpening Based on Intrinsic Image Decomposition

Xudong Kang, Shutao Li, Leyuan Fang, Jon Atli Benediktsson

Research output: Contribution to journalArticlepeer-review

3 Citations (Scopus)


A fused image of high spatial and spectral resolutions can be obtained by fusing a panchromatic (PAN) image with a multi-spectral (MS) image. In this paper, a new image fusion method is proposed, based on an intrinsic image decomposition model which assumes that an image can be separated into two components: the reflectance and illumination components. In pansharpening, it is known that the PAN image is a good substitute for the illumination of the ideal high resolution MS image. Therefore, the reflectance of the low resolution MS image can be estimated with the MS image and the downsampled PAN image. Then, through combining the upsampled reflectance component with the high resolution illumination component (the original PAN image), the pansharpened high resolution MS image can be reconstructed. Experiments performed on three data sets captured by different satellite sensors demonstrate that the proposed method can obtain clear fused images without causing a serious spectral distortion. Furthermore, since the proposed method requires dot multiplication, division, downsampling, and upsampling operations to be performed only once, it can be implemented for a very fast performance.

Original languageEnglish
Article number94
JournalSensing and Imaging
Issue number1
Publication statusPublished - Nov 2014

Bibliographical note

Funding Information:
Acknowledgments The authors would like to thank S. Rahmani and J. Choi for providing the software of the Brovey, AIHS, and CSPR methods. This paper was supported in part by the National Natural Science Foundation for Distinguished Young Scholars of China under Grant No. 61325007, the National Natural Science Foundation of China under Grant No. 61172161.

Publisher Copyright:
© 2014, Springer Science+Business Media New York.

Other keywords

  • Image fusion
  • Intrinsic image decomposition
  • Pansharpening
  • Spectral distortion


Dive into the research topics of 'Pansharpening Based on Intrinsic Image Decomposition'. Together they form a unique fingerprint.

Cite this