Enhancing Low-Resolution Persian License Plates via Diffusion Models

Document Type : Research Article

Authors

1 Department of Telecommunications and Electronics Engineering, Faculty of Electrical and Computer Engineering, Graduate University of Advanced Technology, Kerman, Iran.

2 Department of Applied Mathematics, Faculty of Modern Sciences and Technologies, Graduate University of Advanced Technology, Kerman, Iran.

10.22108/jcs.2026.145412.1171

Abstract

With the rapid growth of urbanization and the increasing number of vehicles, automatic license plate recognition systems have become an essential component of traffic monitoring and management. However, the accuracy of these systems decreases when the captured plate images have low resolution or are affected by environmental noise. This study investigates the use of diffusion models to enhance the resolution of Iranian license plate images under such challenging conditions. In the proposed framework, a diffusion-based denoising process progressively reconstructs and refines image details iteratively. Experimental evaluations demonstrate that the proposed method outperforms well-known approaches such as Bicubic interpolation and deep learning-based super-resolution models, including SRGAN, Real-ESRGAN, BSRGAN, and SwinIR, in terms of perceptual quality and objective metrics. Specifically, the model achieves a PSNR of 31.2458 and an SSIM of 0.8621. Moreover, OCR results indicate that the reconstructed images produced by the diffusion model lead to better character readability and enhanced recognition accuracy.

Keywords

Main Subjects

References

[1] Sikora, P. and Malina, L. and Kiac, M. and Martinasek, Z. and Riha, K. and Prinosil, J. and Jirik, L. and Srivastava, G.. Artificial intelligence-based surveillance system for railway crossing traffic. IEEE Sensors Journal. 21(14): 15515-15526, 2021. [DOI ]
[2] Dong, C. and Loy, C. C. and He, K. and Tang, X.. Image super-resolution using deep convolutional networks. IEEE Transactions on Pattern Analysis and Machine Intelligence. 38(2): 295-307, 2016. [DOI ]
[3] Ledig, C. and Theis, L. and Huszar, F. and Caballero, J. and Cunningham, A. and Acosta, A. and Aitken, A. P. and Tejani, A. and Totz, J. and Wang, Z. and Shi, W.. Photo-realistic single image super-resolution using a generative adversarial network. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). 4681-4690, 2017. [DOI ]
[4] Wang, X. and Yu, K. and Wu, S. and Gu, J. and Liu, Y. and Dong, C. and Qiao, Y. and Loy, C. C.. ESRGAN: Enhanced super-resolution generative adversarial networks. Computer Vision - ECCV 2018 Workshops. 11133: 63-79, Springer. 2019. [DOI ]
[5] Dhariwal, P. and Nichol, A.. Diffusion models beat GANs on image synthesis. Advances in Neural Information Processing Systems (NeurIPS). 34: 8780-8794, 2021. [DOI ]
[6] Ho, J. and Jain, A. and Abbeel, P.. Denoising diffusion probabilistic models. Advances in Neural Information Processing Systems (NeurIPS). 33: 6840-6851, 2020. [DOI ]
[7] Saharia, C. and Ho, J. and Chan, W. and Salimans, T. and Fleet, D. J. and Norouzi, M.. Image super-resolution via iterative refinement. IEEE Transactions on Pattern Analysis and Machine Intelligence. 45(4): 4713-4726, 2023. [DOI ]
[8] Saharia, C. and Chan, W. and Chang, H. and Lee, C. and Ho, J. and Salimans, T. and Fleet, D. J. and Norouzi, M.. Palette: Image-to-image diffusion models. ACM SIGGRAPH 2022 Conference Proceedings. 1-10, 2022. [DOI ]
[9] Rahmani, M. and Sabaghian, M. and Moghadami, S. M. and Talaie, M. M. and Naghibi, M. and Keyvanrad, M. A.. {IR-LPR}: A large scale Iranian license plate recognition dataset. 2022 12th International Conference on Computer and Knowledge Engineering (ICCKE). 53-58, IEEE. 2022. [DOI ]
[10] Keys, R.. Cubic convolution interpolation for digital image processing. IEEE Transactions on Acoustics, Speech, and Signal Processing. 29(6): 1153-1160, 1981. [DOI ]
[11] Wang, X. and Xie, L. and Dong, C. and Shan, Y.. {Real-ESRGAN}: Training real-world blind super-resolution with pure synthetic data. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) Workshops. 1905-1914, 2021. [DOI ]
[12] Zhang, K. and Liang, J. and Van Gool, L. and Timofte, R.. Designing a practical degradation model for deep blind image super-resolution. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). 4791-4800, 2021. [DOI ]
[13] Liang, J. and Cao, J. and Sun, G. and Zhang, K. and Van Gool, L. and Timofte, R.. SwinIR: Image restoration using Swin transformer. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) Workshops. 1833-1844, 2021. [DOI ]
[14] Sun, J. and Xu, Z. and Shum, H.-Y.. Image super-resolution using gradient profile prior. IEEE Conference on Computer Vision and Pattern Recognition (CVPR). 1-8, IEEE. 2008. [DOI ]
[15] Freedman, G. and Fattal, R.. Image and video upscaling from local self-examples. ACM Transactions on Graphics (TOG). 30(2): 12:1-12:11, 2011. [DOI ]
[16] Kim, K. I. and Kwon, Y.. Single-image super-resolution using sparse regression and natural image prior. IEEE Transactions on Pattern Analysis and Machine Intelligence. 32(6): 1127-1133, 2010. [DOI ]
[17] Chang, H. and Yeung, D.-Y. and Xiong, Y.. Super-resolution through neighbor embedding. Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR). 1: 275-282, IEEE. 2004. [DOI ]
[18] Yang, J. and Wright, J. and Huang, T. S. and Ma, Y.. Image super-resolution via sparse representation. IEEE Transactions on Image Processing. 19(11): 2861-2873, 2010. [DOI ]
[19] Dong, C. and Loy, C. C. and Tang, X.. Accelerating the super-resolution convolutional neural network. Computer Vision - ECCV 2016. 391-407, Springer. 2016. [DOI ]
[20] Shi, W. and Caballero, J. and Huszar, F. and Totz, J. and Aitken, A. P. and Bishop, R. and Rueckert, D. and Wang, Z.. Real-time single image and video super-resolution using an efficient sub-pixel convolutional neural network. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). 1874-1883, 2016. [DOI ]
[21] Frolov, S. and Hinz, T. and Raue, F. and Hees, J. and Dengel, A.. Adversarial text-to-image synthesis: A review. Neural Networks. 144: 187-209, 2021. [DOI ]
[22] Moser, B. B. and Shanbhag, A. S. and Raue, F. and Frolov, S. and Palacio, S. and Dengel, A.. Diffusion models, image super-resolution, and everything: A survey. IEEE Transactions on Neural Networks and Learning Systems. 36(7): 11793-11813, 2025. [DOI ]
[23] Song, Y. and Ermon, S.. Generative modeling by estimating gradients of the data distribution. Advances in Neural Information Processing Systems (NeurIPS). 32: 11918-11930, 2019. [DOI ]
[24] Song, Y. and Sohl-Dickstein, J. and Kingma, D. P. and Kumar, A. and Ermon, S. and Poole, B.. Score-based generative modeling through stochastic differential equations. International Conference on Learning Representations (ICLR). 2021. [DOI ]
[25] Wang, Y. and Yang, W. and Chen, X. and Wang, Y. and Guo, L. and Chau, L.-P. and Liu, Z. and Qiao, Y. and Kot, A. C. and Wen, B.. {SinSR}: Diffusion-based image super-resolution in a single step. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). 25796-25805, 2024. [DOI ]
[26] Yue, Z. and Wang, J. and Loy, C. C.. ResShift: Efficient diffusion model for image super-resolution by residual shifting. Advances in Neural Information Processing Systems (NeurIPS). 36: 13294-13307, 2023. [DOI ]
[27] Yang, Y. and Bi, P. and Liu, Y.. License plate image super-resolution based on convolutional neural network. 2018 IEEE 3rd International Conference on Image, Vision and Computing (ICIVC). 723-727, IEEE. 2018. [DOI ]
[28] Shah, B. K. and Yadav, A. and Dixit, A. K.. License plate image super-resolution using generative adversarial network (GAN). 2022 International Conference on Applied Artificial Intelligence and Computing (ICAAIC). 1139-1143, IEEE. 2022. [DOI ]
[29] Mei, Y. and Moelter, M. and Haddad, R. J.. License plate image resolution enhancement using super-resolution generative adversarial network. IEEE SoutheastCon 2024. 1262-1267, IEEE. 2024. [DOI ]
[30] Pan, Y. and Tang, J. and Tjahjadi, T.. LPSRGAN: Generative adversarial networks for super-resolution of license plate image. Neurocomputing. 580: 127426, 2024. [DOI ]
[31] Wang, C.-H.. Using super-resolution imaging for recognition of low-resolution blurred license plates: A comparative study of Real-ESRGAN, A-ESRGAN, and StarSRGAN. arXiv preprint arXiv:2403.15466. 2024. [DOI ]
[32] AlHalawani, S. and Benjdira, B. and Ammar, A. and Koubaa, A. and Ali, A. M.. DiffPlate: A diffusion model for super-resolution of license plate images. Electronics. 13(13): 2670, 2024. [DOI ]
[33] Wang, Z. and Bovik, A. C. and Sheikh, H. R. and Simoncelli, E. P.. Image quality assessment: from error visibility to structural similarity. IEEE Transactions on Image Processing. 13(4): 600-612, 2004. [DOI ]
[34] Wang, Z. and Simoncelli, E. P. and Bovik, A. C.. Multiscale structural similarity for image quality assessment. The Thirty-Seventh Asilomar Conference on Signals, Systems & Computers. 2: 1398-1402, IEEE. 2003. [DOI ]
[35] Afkari-Fahandari, A. and Shabaninia, E. and Asadi-Zeydabadi, F. and Nezamabadi-Pour, H.. A comprehensive survey of transformers in text recognition: Techniques, challenges, and future directions. ACM Computing Surveys. 58(5): 111:1-111:42, 2026. [DOI ]
[36] Asadi-Zeydabadi, F. and Afkari-Fahandari, A. and Faraji, A. and Shabaninia, E. and Nezamabadi-Pour, H.. IDPL-PFOD2: A new large-scale dataset for printed Farsi optical character recognition. arXiv preprint arXiv:2312.01177. 2023. [DOI ]
[37] Asadi-Zeydabadi, F. and Shabaninia, E. and Nezamabadi-Pour, H.. Enhancing license plate recognition using a language model-based approach. Journal of Machine Vision and Image Processing. 11(4): 15-26, 2025. [DOI ]

Files

History

  • Receive Date: 26 May 2025
  • Revise Date: 04 July 2026
  • Accept Date: 15 August 2026

Share

How to cite

Statistics