In: Wagner W., Székely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Voi. XXXVIII, Part 7B
3) Location models improvement using a space-triangulation
process and model parameters correction, including a possible
DEM refining.
Moreover, thanks to the high agility and the precise pointing
capability of the Pleiades-HR platform, it will be possible to
acquire several successive adjacent strips within a unique pass
over a targeted area.
Thus, the Pleiades-HR ortho-mosaic product will provide
end-users with an image of larger size, as being the
result of a seamless patchwork of individual strips. The
output coverage may reach up to 10 000 km 2 and even
more, according to the length of the neighbouring strips
and the tolerated B/H ratio between acquisitions.
Figure 5: Pleiades-HR agility and mosaicking capability
The automatic mosaicking process relies on two geometric
refining phases (as described above). The same method is used
for:
Phase 1: Absolute refining of each data strip location model
using Reference3D™ data.
Phase 2: Relative and finer improvement of location models
using overlapping areas of adjacent strips (350m to 3.5km) and
Reference3D™ DEM. The evaluation has been done using
aerial images simulating Pleiades-HR acquisitions and the
results were excellent. Thanks to the first phase, the absolute
accuracy is less than 2.5m (0.5 Reference3D™ pixels). The
second phase refines the relative model with respect to the very
high resolution of the acquisition (<0.5 pixels HR).
This product represents the successful realization of the
Pleiades-HR mission since it takes advantage of the main
characteristics of the system: agility and high location
precision. The automatic mosaicking process relies on
automatic tie point selection (Baillarin 2004) and geometric
models refining phases followed by local radiometric
homogenisation and stitching phases (seam-less line
computation).
Figure 6: Pleiades-HR mosaic (and stitching line)
5. CONCLUSION AND OUTLOOK
The Pleiades-HR image processing is very complex (about 50
times more complex than SPOT5) and therefore time
consuming, especially for geometric corrections. Nevertheless,
a ground processing unit has been successfully developed to
handle up to an average of 200 products a day for the main
civilian UGC in Spot Image premises. Based on a multi
clusters architecture (IBM Blade Center™) linked by high
speed Fibre Channel to a Storage Area Network, the Pleiades-
HR infrastructure is highly scalable for the different centres
needs (including local receiving stations). In particular, for
emergency needs, pan-sharpened ortho-rectified products
(20x20km 2 ) are processed in less than 45mn and mosaic
products (60x60km 2 ) in less than 140mn.
The realization of the Pleiades-HR ground segment is once
more the opportunity to demonstrate that a global optimization
of the system from board to ground design must be considered
in order to relax requirements of on-board architecture and
lower the overall cost of the mission.
REFERENCES
Baillarin S, 2006, Remote Sensing Image Ground Segment
Interoperability: PLEIADES-HR case study, Proceedings
IGARSS, Denver, 2006
Kubik P, 2005, PLEIADES image quality from users’ needs to
products definition, SPIE Europe International Symposium on
Remote Sensing, Bruges, Belgium, September 19-22, 2005
Unser M, 1999, Splines: A perfect fit for signal and image
processing, IEEE Signal Processing Magazine, November
1999.
De Lussy F, 2006, Pleiades-HR image system products and
geometric accuracy, ISPRS, Hanover, 2006
Tao C., Hu Y., 2001, The Rational Function Model: a tool for
processing high resolution imagery. Earth Observation
Magazine, 2001