version française rss feed
HAL : hal-00687505, version 1

Fiche détaillée  Récupérer au format
PIERS 2012, Kuala Lumpur : Malaisie (2012)
Target detection based on morphological component analysis of HFSWR images for maritime surveillance
Alexandre Baussard 1, 2, Samuel Grosdidier 3

To supplement actual systems (like the Automatic Identification System - AIS) to monitor the maritime traffic in given areas within the Exclusive Economic Zone (EEZ - 200nm), High Fre- quency Surface Wave (HFSW) radar seem to be good candidates. Indeed recent works [1, 2] show that they can provide useful informations even if the spatial and temporal resolutions are weaks. A global surveillance system of the EEZ and harbor could also combine HFSW radar for the long range detection and X-band radar for short ranges (<30km). These last systems are more suitable for near area, which are more vulnerable, since they provide good spatial and temporal resolutions. HFSW radars have been efficiently used these last three decades to remotely measure oceano- graphic parameters. They can provide surface currents, wave spectra, wind intensities and di- rections. In this contribution, as already introduced, these systems are considered for traffic surveillance. The used system is a WEllen RAdar (WERA), with a central frequency between 12 and 13MHz, which provides Range-Doppler (RD) images to be processed for the detection part. These RD images, for the detection purpose, are strongly polluted by sea clutter and other interference leading to a challenging problem. In the paper, a short review of previous works about simulating HFSW images [3] and the validation against real data will be given. Then the image processing approach will be detailed. The proposed method is based on the Morphological Component Analysis (MCA) [5, 6]. Due to HF image features, the method had required some adaptations and some first results have been already given in a previous work [4]. In this contribution, some new modifications of the MCA algorithm are introduced in order to enhance the extraction of the target signatures. The new algorithm includes some processing during the iterative MCA process against the multiscale coefficients (this will be detailed in the full paper). Some results from simulated and real data will be display to illustrate the efficiency of the proposed method.
1 :  Pôle STIC [Brest] (STIC)
ENSTA Bretagne
2 :  Laboratoire des sciences et techniques de l'information, de la communication et de la connaissance (Lab-STICC)
CNRS : UMR3192 – Université de Bretagne Occidentale (UBO) – Université de Bretagne Sud (UBS) – Institut Mines-Télécom – Télécom Bretagne – PRES Université Européenne de Bretagne (UEB) – Institut Supérieur des Sciences et Technologies de Brest (ISSTB)
3 :  Laboratoire de sondages électromagnétiques de l'environnement terrestre (LSEET)
CNRS : UMR6017 – INSU – Université Sud Toulon Var
Sciences de l'ingénieur/Traitement du signal et de l'image

Informatique/Traitement du signal et de l'image

Sciences de l'ingénieur/Electromagnétisme
radar – high frequency surface wave