Method and software for image quality enhancement
DOI:
https://doi.org/10.34185/1562-9945-4-165-2026-02Keywords:
image enhancement, adaptive processing, GFPGAN, Real-ESRGAN, AIEA, PSNR, SSIM, digital image processing, neural networksAbstract
Image enhancement remains an important problem because real scenes are captured under different illumination conditions, with varying object positions, blur, noise, and loss of fine details. Classical techniques based on histogram equalization, brightness correction, and filtering are still widely used because of their low computational cost, yet they are often insufficiently adaptive to the scene content. At the same time, deep-learning-based models such as GFPGAN and Real-ESRGAN provide high-quality restoration results for specific categories of images, although their efficiency strongly depends on whether the chosen architecture matches the actual input data.
The purpose of this paper is to develop an adaptive method and software for image quality enhancement that combines the digital AIEA method with the GFPGAN and Real-ESRGAN neural-network models and automatically selects the processing route according to the detected image type.
The proposed method consists of preliminary image analysis, image-type identification, and content-aware routing to a specialized enhancement algorithm. Portrait images are processed by GFPGAN, which restores facial structure and local details. Blurred or low-detail images are processed by Real-ESRGAN, which improves sharpness, detail reproduction, and effective resolution. Other scene types are enhanced by the AIEA digital method, which performs global tonal correction, contrast improvement, and local detail enhancement. The software was implemented as a desktop application that supports automatic routing, manual algorithm selection, result visualization, and saving of enhanced images. Experimental evaluation was carried out on three image classes: portrait images, blurred or low-detail images, and general-scene images. Quantitative comparison using PSNR and SSIM showed that the best results are obtained when the selected algorithm matches the image content: GFPGAN is the most effective for portraits, Real-ESRGAN is the most effective for blurred or low-detail images, and AIEA provides balanced enhancement for general-scene images.
The proposed approach adapts the enhancement route to the image content and combines the advantages of classical digital processing with specialized neural-network models. This improves the flexibility of the software and provides a practical basis for further extension with new processing routes and more accurate image-type identification.
References
Ledig, C., Theis, L., Huszár, F., Caballero, J., Cunningham, A., Acosta, A., Aitken, A., Te-jani, A., Totz, J., Wang, Z., & Shi, W. (2017). Photo-realistic single image super-resolution using a generative adversarial network. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (pp. 105–114). https://doi.org/10.1109/CVPR.2017.19
Zhang, K., Zuo, W., Chen, Y., Meng, D., & Zhang, L. (2017). Beyond a Gaussian de-noiser: Residual learning of deep CNN for image denoising. IEEE Transactions on Image Processing, 26(7), 3142–3155. https://doi.org/10.1109/TIP.2017.2662206
Lim, B., Son, S., Kim, H., Nah, S., & Lee, K. M. (2017). Enhanced deep residual networks for single image super-resolution. In Proceedings of the IEEE Conference on Computer Vi-sion and Pattern Recognition Workshops (pp. 1132–1140). https://doi.org/10.1109/CVPRW.2017.151
Wang, X., Yu, K., Wu, S., Gu, J., Liu, Y., Dong, C., Loy, C. C., Qiao, Y., & Tang, X. (2018). ESRGAN: Enhanced super-resolution generative adversarial networks. ECCV Work-shops. https://doi.org/10.48550/arXiv.1809.00219
Shocher, A., Cohen, N., & Irani, M. (2018). Zero-shot super-resolution using deep internal learning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Rec-ognition (pp. 3118–3126). https://doi.org/10.1109/CVPR.2018.00329
Wang, Z., Bovik, A. C., Sheikh, H. R., & Simoncelli, E. P. (2004). Image quality assess-ment: From error visibility to structural similarity. IEEE Transactions on Image Processing, 13(4), 600–612. https://doi.org/10.1109/TIP.2003.819861
Horé, A., & Ziou, D. (2010). Image quality metrics: PSNR vs. SSIM. In 2010 20th Interna-tional Conference on Pattern Recognition (pp. 2366–2369). https://doi.org/10.1109/ICPR.2010.579
OpenCV. (2026). OpenCV official documentation. Retrieved April 2, 2026, from https://docs.opencv.org/4.x/
Gonzalez, R. C., & Woods, R. E. (2018). Digital image processing (4th ed.). Pearson.
Wang, X., Li, Y., Zhang, H., & Shan, Y. (2021). Towards real-world blind face restora-tion with generative facial prior. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 9168–9178). https://doi.org/10.1109/CVPR46437.2021.00905
Wang, X., Xie, L., Dong, C., & Shan, Y. (2021). Real-ESRGAN: Training real-world blind super-resolution with pure synthetic data. In Proceedings of the IEEE/CVF International Conference on Computer Vision Workshops (pp. 1905–1914). https://doi.org/10.1109/ICCVW54120.2021.00217
Downloads
Published
Issue
Section
License
Copyright (c) 2026 System technologies
This work is licensed under a Creative Commons Attribution 4.0 International License.
