Skip to Main content Skip to Navigation
Theses

Traitement des signaux pour la détection de mines antipersonnel

Abstract : The millions of landmines spread out over the planet are not only a humanitarian disaster, they also hinder the social and economic development of the concerned countries. In this thesis, new signal processing methods are proposed for the detection and localization of landmines on data recorded by a GPR (Ground Penetrating Radar). First, two digital filters are designed in order to remove clutter in Bscan and Cscan data delivered by GPRs. These two kinds of data are respectively interpreted as vertical and horizontal slices of the ground. In order to design the digital filter a frequency analysis of a clutter geometrical model and a geometrical model of a signal coming from a landmine is led for each type of data. Second, a new detection method, based on a nonparametric abrupt changes detection technique, is proposed in order to detect and localize landmines in Bscan data. The method consists in searching spatial abrupt changes in order to detect the possible horizontal landmines position and in searching time abrupt changes in order to detect the buried objects response times. A detection method, based on contours extraction, is also proposed in order to automatically localize landmines in Cscan data. The performances of these two landmines detection method are studied in terms of detection probability and false alarm probability.
Complete list of metadatas

Cited literature [61 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-00180169
Contributor : Delphine Potin <>
Submitted on : Wednesday, October 17, 2007 - 8:34:56 PM
Last modification on : Tuesday, November 24, 2020 - 2:18:22 PM
Long-term archiving on: : Monday, September 24, 2012 - 1:40:25 PM

Identifiers

  • HAL Id : tel-00180169, version 1

Collections

Citation

Delphine Potin. Traitement des signaux pour la détection de mines antipersonnel. Traitement du signal et de l'image [eess.SP]. Ecole Centrale de Lille, 2007. Français. ⟨tel-00180169⟩

Share

Metrics

Record views

703

Files downloads

5631