ANR PHOENIX
Workgroup STAMPIONS
[Space Time Adaptive Multivariate
Processing of Image Observations and Non-Stationarities]
[In collaboration with ONERA and Supéléc/SONDRA]
Context:
Remote sensing data with passive (multispectral,
hyperspectral, etc.) or active (synthetic aperture radar, etc.) sensors offer a
unique opportunity to record, to analyze and to predict the evolution of the
earth, city, and region of interest. In the last decade, many new satellite
remote sensing missions (Sentinel-1, forthcoming Copernicus program) have been
launched, resulting in drastic improvement in the image acquisition
capabilities. With regular acquisition plans and free data access policy. This
results in new challenge for handling and processing such huge volume of data.
This increasing number of Earth Observation systems involves an enhanced
possibility to acquire multi-temporal and multi-dimensional images of the Earth
surface, with improved temporal and spatial resolution. Such new scenario
significantly increases the interest of the time series processing. The
development of novel data processing techniques to address new important and
challenging applications seems promising.
The potential applications are abundant:
infrastructure monitoring, long time sea or urban traffic surveillance.
Objectives:
The objective of ‘STAMPIONS’ is to develop new
methodologies related to multi-temporal data analysis, by investigating:
·
Adaptive
detection from parametric multivariate random vector models that tolerate mismatches,
in a context of robust statistics.
·
Using
the observable spatial structure in order to take into account directionality
and multi-resolution information.
·
Reducing
the computational cost by obtaining a parsimonious description of image time
series.
·
Application
on real SAR data (from Sentinel 1A and 1B) and optic data (from Sentinel 2A and
2B).
Team