We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Electrical Engineering and Systems Science > Image and Video Processing

Title: Model-based image adjustment for a successful pansharpening

Abstract: A new model-based image adjustment for the enhancement of multi-resolution image fusion or pansharpening is proposed. Such image adjustment is needed for most pansharpening methods using panchromatic band and/or intensity image (calculated as a weighted sum of multispectral bands) as an input. Due various reasons, e.g. calibration inaccuracies, usage of different sensors, input images for pansharpening: low resolution multispectral image or more precisely the calculated intensity image and high resolution panchromatic image may differ in values of their physical properties, e.g. radiances or reflectances depending on the processing level. But the same objects/classes in both images should exhibit similar values or more generally similar statistics. Similarity definition will depend on a particular application. For a successful fusion of data from two sensors the energy balance between radiances/reflectances of both sensors should hold. A virtual band is introduced to compensate for total energy disbalance in different sensors. Its estimation consists of several steps: first, weights for individual spectral bands are estimated in a low resolution scale, where both multispectral and panchromatic images (low pass filtered version) are available, then, the estimated virtual band is up-sampled to a high scale and, finally, high resolution panchromatic band is corrected by subtracting virtual band. This corrected panchromatic band is used instead of original panchromatic image in the following pansharpening. It is shown, for example, that the performance quality of component substitution based methods can be increased significantly.
Comments: 17 pages, 6 figures, 7 tables
Subjects: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
Cite as: arXiv:2103.03062 [eess.IV]
  (or arXiv:2103.03062v1 [eess.IV] for this version)

Submission history

From: Gintautas Palubinskas [view email]
[v1] Thu, 4 Mar 2021 14:38:22 GMT (936kb)

Link back to: arXiv, form interface, contact.