Full text loading...
-
Automatic Classification of Metallic, Plastic and Concrete Targets Buried at IAG/USP Geophysical Test Site Using ANN and GPR Methodologies - First Results
- Publisher: European Association of Geoscientists & Engineers
- Source: Conference Proceedings, Near Surface 2011 - 17th EAGE European Meeting of Environmental and Engineering Geophysics, Sep 2011, cp-253-00006
- ISBN: 978-90-73834-15-6
Abstract
A methodology for classifying automatically metallic, plastic and concrete targets using pattern recognition techniques on GPR data under controlled field conditions was developed. The method consists to develop an Artificial Neural Network (ANN) classifier, using the multilayer perceptron (MLP), with features extracted from GPR profiles over targets in subsoil, and then using it to classify diffraction hyperbolas indicating their position and depth. The classification allows a high resolution reconstruction of the subsurface with reduced computing time. The system was developed in MATLAB and applied to data obtained from the IAG-USP test site, located in the city of São Paulo, Brazil, containing metallic and plastic drums and pipes and concrete tubes under controlled field conditions. The results using real data indicate that the automatic classification of the targets in the subsoil is efficient, contributing for ambiguities reduction in the near surface geophysical data interpretation, besides having application on mapping of targets in subsoil.