Home / Current Issue / Paper 1724543
Automated Classification of Subsurface Cracks in GPR B-Scans Using CNN and SVM
Subject area: Science,Engineering and Technology · Area of research: Machine Learning
Abstract
Ground Penetrating Radar (GPR) is widely used for non-destructive evaluation of pavements and concrete structures. However, manual interpretation of GPR B-scan images is time-consuming and subjective. This work proposes a machine learning framework combining Convolutional Neural Networks (CNN) and Support Vector Machines (SVM) for automatic classification of subsurface crack sizes. Synthetic GPR data are generated using gprMax under controlled simulation conditions for different material permittivities and crack geometries. The CNN learns spatial features directly from B-scan images, while the SVM serves as a baseline classifier using handcrafted statistical features. Models are trained on discrete crack depths and evaluated on both seen and unseen datasets to assess generalization. Results show that the CNN significantly outperforms the SVM, especially on unseen crack depths, demonstrating its ability to learn physically meaningful features from GPR data.
Keywords
Convolutional Neural Networks, Crack Detection, Deep Learning, Ground Penetrating Radar, Non-Destructive Testing, Support Vector Machine
References
[1] A. Giannopoulos, “Modelling ground penetrating radar by gprMax,” Construction and Building Materials, 2005.
[2] A. Giannopoulos et al., “gprMax: Open source software to simulate electromagnetic wave propagation for GPR,” Computer Physics Communications, 2016. 10.1016/j.cpc.2016.08.020
[3] C. Cortes and V. Vapnik, “Support-vector networks,” Machine Learning, 1995.
[4] Y. LeCun, Y. Bengio, and G. Hinton, “Deep learning,” Nature, 2015.
[5] I. Goodfellow, Y. Bengio, and A. Courville, Deep Learning. Cambridge, MA, USA: MIT Press, 2016. 10.5555/3086952.
[6] C. M. Bishop, Pattern Recognition and Machine Learning. New York, NY, USA: Springer, 2006. 10.1007/978-0-387-45528-0.
[7] Y. Ma, Y. Wu, and J. Liu, “Automatic identification of subsurface defects using GPR and deep learning,” NDT & E International, 2018.
How to cite this paper
@article{1724543,
author = {Precious Nnamdi Ibe, Mubarak Adekunle Jimoh},
title = {Automated Classification of Subsurface Cracks in GPR B-Scans Using CNN and SVM},
journal = {Iconic Research And Engineering Journals},
year = {2026},
volume = {9},
number = {9},
pages = {4142-4145},
issn = {2456-8880},
url = {https://www.irejournals.com/formatedpaper/1724543.pdf},
abstract = {Ground Penetrating Radar (GPR) is widely used for non-destructive evaluation of pavements and concrete structures. However, manual interpretation of GPR B-scan images is time-consuming and subjective. This work proposes a machine learning framework combining Convolutional Neural Networks (CNN) and Support Vector Machines (SVM) for automatic classification of subsurface crack sizes. Synthetic GPR data are generated using gprMax under controlled simulation conditions for different material permittivities and crack geometries. The CNN learns spatial features directly from B-scan images, while the SVM serves as a baseline classifier using handcrafted statistical features. Models are trained on discrete crack depths and evaluated on both seen and unseen datasets to assess generalization. Results show that the CNN significantly outperforms the SVM, especially on unseen crack depths, demonstrating its ability to learn physically meaningful features from GPR data.},
keywords = {Convolutional Neural Networks, Crack Detection, Deep Learning, Ground Penetrating Radar, Non-Destructive Testing, Support Vector Machine},
month = {March},
}