This paper presents a method for target-specific feature extraction from ground penetrating radar (GPR) signatures and examines the applicability of such features to classification of minelike targets. The signatures of a set of targets measured in soils with different dielectric permitivities and at various burial depths are used in the experiments. Each target, buried in one soil and at one particular depth, is represented by a set of complex poles computed from the ensemble of neighbouring target-specific signatures. The empirical probability density of the corresponding set of pole frequencies is modelled as a mixture of univariate Gaussian functions. Robust partial modelling algorithm is applied to determine the number of Gaussians in the mixture and to estimate their parameters. The procedure that uses the resulting target-specific Gaussian mixtures to compute the class-conditional probabilities of a target is applied for target classification.


    Access

    Access via TIB

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    Feature extraction and classification of minelike targets from GPR data using Gaussian mixture models


    Contributors:
    Carevic, D. (author) / Chant, I. (author) / Caelli, T. (author)


    Publication date :

    1999


    Size :

    6 Seiten, 10 Quellen




    Type of media :

    Conference paper


    Type of material :

    Print


    Language :

    English




    Simultaneous high-dimensional clustering and feature selection using asymmetric Gaussian mixture models

    Elguebaly, T. / Bouguila, N. | British Library Online Contents | 2015


    Gaussian Mixture Models for Parking Demand Data

    Fiez, Tanner / Ratliff, Lillian J. | IEEE | 2020



    Data-Driven Remaining Useful Life Estimation Using Gaussian Mixture Models

    Liu, Yixuan / Hu, Zhen / Todd, Michael et al. | AIAA | 2021


    Gaussian-Mixture Probability Hypothesis Density Filter for Multiple Extended Targets

    Han, Y. / Zhu, H. / Han, C. et al. | British Library Online Contents | 2014