Remote sensing image fusion using multiscale mapped LS-SVM

Sheng Zheng, Wen Zhong Shi, Jian Liu, Jinwen Tian

Research output: Journal article publicationJournal articleAcademic researchpeer-review

118 Citations (Scopus)


The panchromatic (Pan) sharpening of multispectral (MS) bands is an important technique in the various applications of satellite remote sensing. This paper presents an MS Pansharpening method using the proposed multiscale mapped least-squares support vector machine (LS-SVM). Under the LS-SVM framework, the salient features underlying the image are represented by support values, and the support value transform (SVT) is developed for image information extraction. The low-resolution MS bands are resampled to the fine scale of the Pan image and sharpened by injecting the detailed features extracted from the high-resolution Pan image. The support value analysis is implemented by using a series of multiscale support value filters that are deduced from the mapped LS-SVM with multiscale Gaussian radial basis function kernels. Experiments are carried out on very high resolution QuickBird MS + Pan data. Fusion simulations on spatially degraded data, whose original MS bands are available for reference, show that the proposed MS Pan-sharpening method performs comparable to the state-of-the-art in terms of the pertained quantitative quality evaluation indexes, such as the Spectral Angle Mapper, relative dimensionless global error in synthesis (ERGAS), modulation-transfer-function-based tool and quality index (Q4), etc. The SVT is an effective tool for remote sensing image fusion.
Original languageEnglish
Pages (from-to)1313-1322
Number of pages10
JournalIEEE Transactions on Geoscience and Remote Sensing
Issue number5
Publication statusPublished - 1 May 2008


  • Image fusion
  • Mapped least-squares support vector machine (mapped LS-SVM)
  • Multiscale Gaussian radial basis functions (RBF)
  • Multispectral (MS) imagery
  • Remote sensing
  • Support value transform (SVT)

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Earth and Planetary Sciences(all)


Dive into the research topics of 'Remote sensing image fusion using multiscale mapped LS-SVM'. Together they form a unique fingerprint.

Cite this