Skip to Main content Skip to Navigation
Theses

Approche parcimonieuse pour l’imagerie 3D haute résolution de surface équivalente radar.

Abstract : The RCS (Radar Cross Section) is a quantity which characterizes the scattering power ofa target exposed to an electromagnetic field. Its analysis and control are important in manyapplications. 3D imaging is a suitable tool to accurately locate and characterize in 3D themain contributors to the RCS. However, this is a non-invertible Fourier synthesis problembecause the number of unknowns is larger than the number of data. Conventional methodssuch as the Polar Format Algorithm, which consists of data reformatting including zeropaddingfollowed by a fast inverse Fourier transform, provide results of limited quality.In this work, we propose a new high resolution method, named SPRITE (for SParse RadarImaging TEchnique), which considerably increases the quality of the estimated RCS maps. Itis based on a regularization scheme that accounts for information of sparsity and support. Thesolution is then defined as the minimizer of a penalized and constrained criterion. Optimizationis ensured by an appropriate adaptation of the ADMM (Alternating Direction Methodof Multiplier) algorithm that is able to quickly perform calculations using fast Fourier transforms.Finally, the method is evaluated on both simulated and real data. Compared to the conventionalmethod, the resolution is significantly increased and the images can support a betterRCS analysis and control.
Complete list of metadatas

https://tel.archives-ouvertes.fr/tel-01944454
Contributor : Abes Star :  Contact
Submitted on : Tuesday, December 4, 2018 - 4:19:10 PM
Last modification on : Friday, October 23, 2020 - 4:37:02 PM

File

BENOUDIBA-CAMPANINI_THOMAS_201...
Version validated by the jury (STAR)

Identifiers

  • HAL Id : tel-01944454, version 1

Citation

Thomas Benoudiba-Campanini. Approche parcimonieuse pour l’imagerie 3D haute résolution de surface équivalente radar.. Traitement du signal et de l'image [eess.SP]. Université de Bordeaux, 2018. Français. ⟨NNT : 2018BORD0110⟩. ⟨tel-01944454⟩

Share

Metrics

Record views

326

Files downloads

135