[Insight-users] Post-doc position at CNES Toulouse
Jordi Inglada
jordi.inglada at cnes.fr
Mon Feb 27 09:10:41 EST 2006
Dear all,
We have an open post-doc position in remote sesning image processing at
CNES Toulouse starting between October and December 2006.
Here goes the research subject :
"Multidimensional rank statistics for similarity measures"
Remote sensing images allow for the monitoring of the Earth's surface
evolution with a (spatial, spectral and temporal) resolution and a field
of view which is always improving. Many processing alorithms are
therefore based on statistical caracterizations of multidimensional
observations:
- In multispectral optical images the links and redundancies between the
different channels are seldom exploited.
- In radar imaging the data are complex. Until now, the circular
gaussian model was commonly used. At metric resolutions, the assumptions
of independence and stationnarity of the observations do not hold
anymore. Hence, new models are needed (for instance in subaperture
analysis).
- In POLSAR imaging, the different channels are only modelled up to the
second order (polarimetric coherence, Wishart statistics) while it would
be interesting to go beyond these models in order to process very high
resolution data.
- In change detection analysis, the similarity measures used are based
on the information theory (mutal information, for instance). The
optimizations used for the estimation of those measures are essentially
based on Edgeworth series expansions. The observed dependences could be
characterized by using parametric models (as copulas, for instance)
instead of using expressions which have a high complexity. In the
multi-sensor case, there is no tool allowing the automatic comparison of
an optical and a radar image.
In this work, non conventional statistical models will be used (higher
order statistics, copulas, rank statistics, etc.) in order to evaluate
their pertinence in the operational processing of Earth observation
remote sensing data.
Interested applicants should contact me by e-mail: jordi.inglada at cnes.fr
Thanks in advance.
Jordi INGLADA
--
CNES - DCT/SI/AP - BPI 1219
18, avenue Edouard Belin
31401 Toulouse Cedex 09 - France
Tel. +33.(0)5.61.27.33.97 - Fax. +33.(0)5.61.28.31.09
More information about the Insight-users
mailing list