[Insight-developers] Ph.D. Dissertation funding oportunity at CNES (France)

[Luis Ibanez]

CNES, the French Space Agency, awards doctoral (PhD) research grants
once a year to grantees working in * engineering sciences *(orbital and
space transport systems) and * sciences using space assets *(Universe,
Earth and microgravity sciences). Candidates from universities or
engineering schools must hold a DEA post-graduate diploma (/ Diplôme
d’Etude Approfondie/) from a university or an equivalent qualification
(masters degree, etc.) to be authorized to register for a doctoral
thesis. This qualification and the authorization to register must be
obtained before the start of the research contract. Preference will be
given to candidates under the age of 25 at 31 December of the calendar
year. Research grant contracts run for one year, renewable twice. The
gross yearly remuneration is €20,112 the first year, €21,036 the second
year and €21,948 the final year.

Please find here the description of an open position I am responsible for:

Similarity measures for abrupt change detection on multi-sensor
satellite images.

Satellite imagery allows for the study and monitoring of the evolution
of land surfaces with a systematic repetitivity which cannot be achieved
by other means. In spite of this fact, in the case of natural or
industrial disasters it is often difficult to produce, in a short delay,
damage analysis and impact maps. The issues making this task difficult
may have the following origins : the lack of a dedicated satellite
constellation, the delays between satellite scheduling and image
acquisition, the image processing time. If we analyze the data
availability problem, the experience in the framework of the
International Charter Space and Major Disasters
<http://www.disasterscharter.org>, shows that most of the time it is
difficult to gather a couple of images (before and after the event)
which is easy to exploit. Indeed, if one tries to reduce the response
time for the end user, one cannot afford to wait for the « best »
acquisition which fits the archive data. This means that one has to
compare images of different natures (resolutions, spectral bands,
physics of the measure, etc.). This leads to long information extraction
delays, since the damage evaluation must be manually performed. It is
thus interesting to try to make this evaluation as automatic as possible.

In this thesis we will study and develop mathematical methods and models
allowing to design similarity measures which are able of showing
landscape evolutions from images acquired by different sensors. The main
goal of the work will be to conceive image analysis methods which are
usable in an operational framework. Research work on similarity measures
and change detection has already been conducted at CNES. Other
approaches (feature-based, object-based) have to be studied and
characterized. CNES' partner SERTIT <http://*sertit*.u-strasbg.fr> has
developed a unique expertise in Europe on image analysis in a crisis
framework by using semi-automatic approaches and CAPI. The research work
to be conducted will rely upon results and end user feedback obtained
through real cases.

The research work will follow these main lines :

1. Study of real cases in order to « factor » common elements in order
to propose a generic methodology for the analysis of abrupt changes
observed on a «before-after» image couple.

2. Study of similarity measures, definition of the similarity concept,
mathematical model for « change », multi-sensor transfer functions.

3. Method choice strategy as a function of the event, landscape and
available data.

4. Validation for real cases using ground truth. Performance evaluation.


Contact person: Jordi Inglada <jordi.inglada at cnes.fr>
Research laboratory: CNES Toulouse & LIS Grenoble.

Best regards,

J. 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