ARIANA Research team
Inverse problems in earth monitoring
- Leader : Josiane Zerubia
- Research center(s) : CRI Sophia Antipolis - Méditerranée
- Field : Perception, Cognition, Interaction
- Theme : Vision, Perception and Multimedia Understanding
- Partner(s) : Université Nice - Sophia Antipolis,CNRS
- Collaborator(s) : U. NICE SOPHIA ANTIPOLIS (UNS), CNRS
Team presentationThis project-team aims are using theoretical models, both stochastic and deterministic, to tackle inverse problems related to Earth observation and cartography. In image processing various problems related to Earth observation are inverse problems. They are ill-posed, as defined by Hadamard. Actually, even in favorable cases (for which a solution exists and is unique) problems remain, that are linked to the instability of the solution due to the noise during the observations. The goal is thus to regularize the solution with a priori constraints on the object. This can be done using either a probabilistic approach, for example a Markovian model (the estimation is then performed in a Bayesian way), or a deterministic approach, for example using a variational model (the estimation is performed by the minimisation of functionals).
- Markovian modelling,
- Variational approach,
- Stochastic geometry,
- Parameter estimation.
Earth observation and cartography is a rich domain. The issue is, for instance, to enhance the quality of a noisy or blurred image in order to extract networks such as roads, rivers, railways, or to classify various objects permitting interpretation by an operator. Earth observation problems are not only two dimensional: high-resolution and super-resolution algorithms applied to optical or radar (SAR) images give 3D Digital Elevation Models.
International and industrial relations
AKKA Tech., Thales Alenia Space, Astrium/EADS, BRGM, Cnes, CS, CemapX, DGA, J.A. Dieudonné Laboratory of Uns, ECP, GdR Isis, GdR MSPC, IFN, IGN, INRA, IRD, MBDA, Paris V and VI, Paris EST, Sagem, Telecom Paris Tech.
Bielorussia (UIIP/BAS Minsk), China (LIAMA Beijing), Czech Republic (UTIA/CAS Prag), Great Britain (University of Bristol, University of Cambridge, University of Durham), Hungary (Szeged University, SZTAKI), Ireland (Trinity College Dublin), Israel (Ben Gurion University, HUJI Jerusalem, Tel-Aviv University, Technion, Weizmann Institute), Italy (University of Genoa, University of Naples, CNR Pisa, University of Rome), Mexico (National Autonomous University of Mexico), the Netherlands (CWI Amsterdam), Russia (IIPT/RAS Moscow, Moscow State University), Tunisia (SupCom Tunis, ENS Tunis), USA (Duke University, Florida State University, University of Illinois at Urbana-Champaign, Wisconsin university).
Keywords: Image processing Inverse problems Variational approach Markovian modeling Probabilistic geometry Parameter estimation Earth observation Remote sensing Cartography Image restoration Network extraction