Clutter Suppression in Ground-Penetrating Radar Images Using an Improved Pix2Pix Model
DOI:
https://doi.org/10.54097/fh9pnb90Keywords:
Ablation Experiment, CBAM, Clutter Suppression, Ground Penetrating Radar, MS‑SSIM, Pix2PixAbstract
To address complex clutter in measured ground-penetrating radar (GPR) B-scan images and the difficulty of obtaining paired clutter-free target-response ground-truth data, this paper proposes an improved Pix2Pix clutter suppression method based on a paired dataset fused by simulated target responses and measured backgrounds. In terms of network structure, residual dense blocks (RDB) are introduced to strengthen feature transmission and reuse. A Convolutional Block Attention Module (CBAM) is adopted to adaptively recalibrate channel‑wise and spatial features. A multi‑scale structural similarity (MS‑SSIM) constraint is added to the generator objective function to improve the structural integrity and continuity of target hyperbolic responses. To control confounding effects caused by weight differences of the L1 loss, comparative experiments are carried out under λL₁ = 100 and λL₁ = 50 respectively. The experimental results show that under λL₁ = 100, the CBAM–MS-SSIM model, Exp3 achieves the best performance within the current data partition and parameter range. Its test‑set PSNR, SSIM and MS‑SSIM reach 52.4344 dB, 0.998056 and 0.998774 respectively. Further ablation experiments indicate that introducing either CBAM or MS‑SSIM alone does not improve performance. When the two modules are combined under a strong L1 reconstruction constraint, they improve target reconstruction, structure preservation, and clutter suppression.
Downloads
References
[1] Zhao, Y., Yang, X., Qu, X., Lan, T., & Gong, J. (2023). Clutter removal method for GPR based on low rank and sparse decomposition with total variation regularization. IEEE Geoscience and Remote Sensing Letters, 20, 3502605, 1–5. https://doi.org/10.1109/LGRS.2023.3250717. DOI: https://doi.org/10.1109/LGRS.2023.3250717
[2] Isola, P., Zhu, J.-Y., Zhou, T., & Efros, A. A. (2017). Image-to-image translation with conditional adversarial networks. arXiv preprint arXiv:1611.07004. https://doi.org/10. 48550/ arXiv. 1611.07004.
[3] Zhang, Y., Tian, Y., Kong, Y., Zhong, B., & Fu, Y. (2018). Residual dense network for image super-resolution. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). DOI: https://doi.org/10.1109/CVPR.2018.00262
[4] Huang, X., Liu, J., Yang, F., Qiao, X., Gao, L., Fu, T., & Zhao, J. (2025). Study on GPR image restoration for urban complex road surfaces using an improved CycleGAN. Remote Sensing, 17(5), 823. https://doi.org/10.3390/rs17050823. DOI: https://doi.org/10.3390/rs17050823
[5] Ni, Z.-K., Shi, C., Pan, J., Zheng, Z., Ye, S., & Fang, G. (2022). Declutter-GAN: GPR B-scan data clutter removal using conditional generative adversarial nets. IEEE Geoscience and Remote Sensing Letters, 19, 4023105, 1–5. https://doi.org/10. 1109/ LGRS.2022.3159788. DOI: https://doi.org/10.1109/LGRS.2022.3159788
[6] Woo, S., Park, J., Lee, J.-Y., & Kweon, I. S. (2018). CBAM: Convolutional block attention module. In Proceedings of the European Conference on Computer Vision (ECCV). DOI: https://doi.org/10.1007/978-3-030-01234-2_1
[7] Wang, Z., Simoncelli, E. P., & Bovik, A. C. (2003). Multiscale structural similarity for image quality assessment. In Proceedings of the 37th Asilomar Conference on Signals, Systems and Computers (Vol. 2, pp. 1398–1402). https://doi. org/ 10. 1109/ACSSC.2003.1292216. DOI: https://doi.org/10.1109/ACSSC.2003.1292216
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Frontiers in Computing and Intelligent Systems
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
