Detection of Buried Landmines using a Convolutional Autoencoder trained on Simulated prompt Gamma Spectra

Authors

  • Konstantinos KARAFASOULIS

DOI:

https://doi.org/10.53477/2284-9378-25-19

Keywords:

Landmine Detection, Artificial Intelligence, Autoencoders, Anomaly Detection, Neutron Activation, Gamma Radiation.

Abstract

The detection of buried landmines remains a persistent challenge in security and humanitarian demining. In this work, we present an indirect detection methodology based on the analysis of prompt gamma-ray emissions induced by 14 MeV neutron irradiation. A high-resolution LaBr₃ detector captures the gamma spectra arising from neutron interactions with soil constituents and buried explosives. A Convolutional Neural Network (CNN) autoencoder, trained in an unsupervised manner, models the intrinsic spectral response of soil under varying moisture conditions. Anomalies between reconstructed and measured spectra are used to infer the presence of subsurface anomalies consistent with landmines. Monte Carlo simulations, conducted with the Geant4 toolkit, generate a comprehensive dataset encompassing a soil matrix under various moisture levels. The proposed system demonstrates sensitivity to buried antipersonnel landmines at shallow depths, validating the integration of neutron activation analysis and deep learning for advanced landmine detection applications.

Author Biography

Konstantinos KARAFASOULIS

Dr Konstantinos Karafasoulis is a member of the Laboratory Teaching Staff at the Hellenic Army Academy and an Associated Researcher at the Aristotle University of Thessaloniki. He earned his BSc in Physics from the Aristotle University (1992) of Thessaloniki and a Ph.D. in High Energy Physics (DELPHI/CERN) from the National Technical University of Athens (1999). With over two decades of expertise, Dr. Karafasoulis specializes in radiation detector simulation, data acquisition systems, and advanced data analysis techniques.

He collaborated on the DELPHI and CMS experiments at CERN, working with teams from Demokritos/Greece and INFN/Italy. Furthermore, Dr. Karafasoulis has contributed to a range of European and National R&D projects focused on refining radiation detectors, data systems, and unique data analysis methods, such as NATO SENERA, FP7-COCAE, EPAN, PYTHAGORAS, PENED, and ESA-C14240. He was the coordinator of the NATO funded project “SENERA: A sensor Network for the localization and Identification of Radiation Sources”.Dr Karafasoulis was involved in the simulation of the MIDAS device, a space radiation dosimeter, using the GEANT4 framework. He also helped develop reconstruction algorithms to identify and measure energy of particles interacting with the device, as part of an (European Space Agency) ESA funded project. He was also responsible for the GEANT4 simulation Work Package of the “Comprehensive radiation monitor package for Lunar mission,” project sponsored by the European Space Agency (ESA).

Currently, he is the secondary Chief Scientific Investigator of the project “Advancing Nuclear Forensics: Machine Learning methods for Nuclear Crime Scene Investgation” funded by the International Atomic Energy Agency (IAEA). In the context of this project he is leading the research for the development of the Machine/Deep Learning algorithms for the detection and identification of radioactive and Special Nuclear Material (SNM).

Dr Karafasoulis is a certified KNIME Data Scientist  and serves as the Academic Supervisor of the course “KNIME: An introduction to data analysis without programming” offered by the Center for Education and LifeLong Learning of the Aristotle University of Thessaloniki, where he enjoys teaching about data analysis.

For a complete list of publications, see: http://karafasoulis.eu/

 

References

Adari, S., and S. Alla. 2024. Beginning Anomaly Detection Using Python-Based Deep.Berkeley: Apress.

Agostinelli, S, Allison J., Amako K., Apostolakis J., Araujo H., Arce P., Asai M., et al. 2003. “Geant4—a simulation toolkit.” Nuclear Instruments and Methods A 506 (3): 250-303. doi:10.1016/S0168-9002(03)01368-8.

Clifford, E.T.H., J.E. McFee, H. Ing, H.R. Andrews, D. Tennant, E. Harper, and A.A. Faust. 2007. “A militarily fielded thermal neutron activation sensor for landmine detection.”Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 579 (1): 418-425. doi:10.1016/j.nima.2007.04.091.

Datema, C.P., V.R. Bom, and C.W.E. van Eijk. 2003. “Monte Carlo simulations of landminedetection using neutron backscattering imaging.” Nuclear Instruments and Methodsin Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 513 (1-2): 398-402.

Elsheikh, N.A.A. 2018. “Monte Carlo modelling of a neutron-induced gamma-ray sensor for landmine or explosive detection.” Journal of Radiation Research and Applied Sciences11 (4): 403–407. https://doi.org/10.1016/j.jrras.2018.08.004.

Gulli, A., and S. Pal. 2017. Deep learning with Keras. Packt Publishing Ltd.

Kingma, Diederik P, and Jimmy Ba. 2015. “Adam: A Method for Stochastic Optimization.”arXiv preprint. doi:10.48550/arXiv.1412.6980.

Landmine and Cluster Munition Monitor. 2024. “Landmine Monitor 2024.” https://www.the-monitor.org/reports/landmine-monitor-2024.

Nebbia, Ginacarlo and Juergen Gerl. 2005. “Detection of buried landmines and hiddenexplosives using neutron, X-ray and gamma-ray probes.” Europhysics News 36: 119-123. doi:10.1051/epn:2005403.

Viesti, G., M. Lunardon, G. Nebbia, M. Barbui, M. Cinausero, G. D’Erasmo, M. Palomba,A. Pantaleo, J. Obhođaš, and V. Valković. 2006. “The detection of landmines byneutron backscattering: Exploring the limits of the technique.” Applied Radiation andIsotopes 64 (6): 706-716. doi:10.1016/j.apradiso.2005.12.017.

Xue, Hui, Guang-Yu Shi, De-Dong He, and Si-Yuan Chen. 2022. “Simulation Study on Landmines Detection by Pulsed Fast Thermal Neutron Analysis.” Arabian Journal for Science and Engineering 47 (1): 879-885. doi:10.1007/s13369-021-05742-0.

Downloads

Published

2025-06-30

How to Cite

KARAFASOULIS, K. . (2025). Detection of Buried Landmines using a Convolutional Autoencoder trained on Simulated prompt Gamma Spectra. BULLETIN OF "CAROL I" NATIONAL DEFENCE UNIVERSITY, 14(2), 114–127. https://doi.org/10.53477/2284-9378-25-19

Issue

Vol. 14 No. 2 (2025)

Section

Articles

License

The published articles are subject to copyright law. All rights are reserved to the “Carol I” National Defense University, regardless of whether all or part of the material are considered, especially the rights to translation, reprinting, re-use of illustrations, quotations, broadcasting through the media, reproduction on microfilms or any other way and storage in data banks. Any replicas without the associated fees are authorized provided the source is acknowledged.