A Machine Learning Approach for Simulating Ground Penetrating Radar

Iraklis Giannakis, Antonios Giannopoulos, Craig Warren

Research output: Chapter in Book/Report/Conference proceedingConference contribution

4 Citations (Scopus)

Abstract

The ability to produce, store and analyse large amounts of well-labeled data as well as recent advancements on supervised training, led machine learning to gain a renewed popularity. In the present paper, the applicability of machine learning to simulate ground penetrating radar (GPR) for high frequency applications is examined. A well-labelled and equally distributed training set is generated synthetically using the finite-difference time-domain (FDTD) method. Special care was taken in order to model the antennas and the soils with sufficient accuracy. Through a stochastic parameterisation, each model is expressed using only seven parameters (i.e. the fractal dimension of water fraction, the height of the antenna and so on). Based on these parameters and the synthetically generated training set, a machine learning framework is trained to predict the resulting A-Scan in real-time. Thus, overcoming the time-consuming calculations required for an equivalent FDTD simulation.
Original languageEnglish
Title of host publication2018 17th International Conference on Ground Penetrating Radar (GPR)
PublisherIEEE Explore
ISBN (Print)978-1-5386-5777-5, 978-1-5386-5778-2
DOIs
Publication statusPublished - 23 Aug 2018
Event17th International Conference on Ground Penetrating Radar -
Duration: 18 Jun 2018 → …

Conference

Conference17th International Conference on Ground Penetrating Radar
Period18/06/18 → …

Fingerprint Dive into the research topics of 'A Machine Learning Approach for Simulating Ground Penetrating Radar'. Together they form a unique fingerprint.

  • Cite this

    Giannakis, I., Giannopoulos, A., & Warren, C. (2018). A Machine Learning Approach for Simulating Ground Penetrating Radar. In 2018 17th International Conference on Ground Penetrating Radar (GPR) IEEE Explore. https://doi.org/10.1109/ICGPR.2018.8441558