PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
October 2016
747
ntroduction
Signal-to-noise ratio (SNR), spectral resolution, and spatial resolution are the three most important elements
of remote sensing imagery for information extraction and information interpretation. However, due to the
physical limitations of optical sensors, it is difficult to acquire an image that has both a high spatial resolution
and a high spectral resolution without sacrificing its SNR.
Therefore, most modern high (spatial) resolution satellites, such as IKONOS, QuickBird, GeoEye, WorldView,
and Pléiades, simultaneously capture two images—a high resolution panchromatic (Pan) image and a low
resolution multispectral (MS) image. This alternative solution overcomes sensor limitations such that both
Pan and MS images have a high SNR. However, the high spatial resolution and high spectral resolution are
associated with two different images.
To combine the high spatial resolution information of the Pan image with the high spectral resolution informa-
tion of the MS image into one image, many pansharpening techniques have been developed. These techniques
fuse the Pan image with the MS image to produce a pansharpened MS image, which is often at the same
spatial resolution as the Pan image. However, it is still an open question as to whether the quality of such a
pansharpened MS image can reach or surpass the quality of the desired high resolution MS image that would
ideally be acquired at the same spatial resolution as the Pan image, because such an ideal high resolution MS
image is not available for quality comparison.
To answer this open question about the quality of the pansharpened image, we performed a research experi-
ment using a standard procedure in the imaging industry. Operating in a controlled environment, a scientific
monochrome charge coupled device (CCD) camera was used to capture high resolution panchromatic images of
three scientific test charts. Scientific color filters were used with the CCD camera to collect high resolution MS
images of the same scene. To achieve high quality illumination of the scene (i.e. the scientific charts), a set of
industrial standard lights was used.
The commercial pansharpening software, Fuze Go (based on UNB PanSharp algorithm used by DigitalGlobe)
was used to fuse the high resolution Pan image with a low resolution MS image, generated by binning the high
resolution MS image. A set of standard software tools
from the imaging industry was used to measure the qual-
ity difference between the original high resolution MS
image and the pansharpened MS image. Details about
image collection, pansharpening, and quality evaluation
are described in the following sections.
U
nderstanding
the
Q
uality
of
P
ansharpening
–
A L
ab
S
tudy
By Yun Zhang, Aditya Roshan, Shabnam Jabari, Sina A. Khiabani,
Fatemeh Fathollahi, and Rakesh K. Mishra
I
Photogrammetric Engineering & Remote Sensing
Vol. 82, No. 10, October 2016, pp. 747–755.
0099-1112/16/747–755
© 2016 American Society for Photogrammetry
and Remote Sensing
doi: 10.14358/PERS.82.10.747
