Objectives

OBJECTIVES :

  • To gather national teams of statistical signal processing in fields linked to RADAR and Image (this is the continuation of Club STAP),
  • explore some scientific breaktroughs and maintain the innovative effort on some thematics for the french defense.

CONTEXTS :

  • Data systems in RADAR and Imagery more and more complex: number of transmitters and/or receivers, multiple diversity (Doppler, Polarisation, Interferometry, temporal axis, spatial structure, transmitted waveforms)
  • Complex background : Heterogeneous clutter, non gaussian, non-stationnary, corrupted learning data, etc.

APPLICATIONS :

  • Radar: surveillance, STAP-GMTI, MIMO (detection of targets with low SER and small speed in a non gaussian and heterogeneous context), mono and multi-look SAR (target detections in SAR, change detection, surveillance of structure) et Goniometry
  • Hyperspectral: detection of outlier, targets and change
  • GPR: detection in complex environments, FOPEN
  • Astronomy: radio-interferometry, calibration of sensors

TECHNICAL ISSUES:

  • Large data size: statistical parameters learning
  • Theoretical performance: false alarms regulation, …
  • Uncertainty on the useful signal: trade-off between hyper-parametrization and estimation accuracy

This project is decomposed in four Works Packages:

  • WP1: Robust estimation for large data size
    • Robust estimation of structured covariance matrix
    • Conjoint estimation of parametrized mean and covariance matrix
    • Implementation of robust algorithms by means of riemannian geometry
  • WP2: Integration of uncertainties on the signal model
    • Bayesian subspace methods for estimation and detection
    • Advanced calibration for multi-sensors systems and study of the impact of the model error on the detection/estimation performance
  • WP3: Theoretical performance study
    • Caracterisation of optimal performance by taking into account of constraints on the physical parameters in non gaussian environnement
    • Theoretical performances of estimators
    • Performance analysis of filterting or detection processes in a large dimension regime
  • WP4: Applications
    • RADAR
    • SAR and hyperspectral images
    • Other applications: GPR, astronomy