(50-3) 09 * << * >> * Russian * English * Content * All Issues

  Full text (PDF)

DOI: 10.18287/COJ1724

Article ID: 1724

Language: English

Abstract:
Modified versions of the Wiener filter that replace the mean with the median (MMWF) can better reduce Gaussian noise in images than the classical Wiener filter (WF). However, performance gradually decreases as the noise variance increases. To overcome these limitations, we propose a modified Wiener filter (MADNWF) that replaces the mean with the scaled median absolute deviation (MADN). Similar to MMWF, our modification to the Wiener filter uses a local kernel to average over each pixel. Instead of using the median alone, we replace it with MADN. We used four public datasets for the experiment: Set12, the Tampere17 noise-free dataset, TID2008, and the BSD68 dataset. The first step of this study was to generate 'degraded' images. To achieve this, we added a random amount of zero-mean Gaussian noise with noise variances ranging from 10 to 90, in increments of 10, to every image in the dataset. For comparison with the WF and MMWF, we then applied the proposed MADNWF to reduce noise in degraded images. We evaluated the resulting performance by comparing the reduced noise images with the original images using the Structural Similarity Index (SSIM) and Peak Signal-to-Noise Ratio (PSNR) metrics. Experimental results demonstrate that MADNWF consistently outperforms the other methods. At a low noise level (variance = 10), MMWF with a 3x3 filter slightly outperforms in fine-structural preservation, achieving a peak SSIM of 0.6676. However, as the noise intensity increases (variances from 20 to 90), MADNWF achieves complete dominance across all datasets. MADNWF achieves the highest PSNR, up to 33.27 dB at low noise levels, representing an improvement of up to 0.28 dB over WF. Under extreme noise conditions (variance = 90), the MADNWF 7x7 configuration achieves superior image cleanliness (up to 28.91 dB), whereas the MADNWF 5x5 setting serves as the optimal trade-off for structure preservation, outperforming MMWF with an SSIM margin increase of up to 0.0276. In conclusion, our results demonstrate that MADNWF helps reduce the noise distribution in natural images.

Keywords:
Gaussian noise reduction, Wiener filter (WF), the scaled median absolute deviation normalized (MADN), median modified Wiener filter (MMWF).

Acknowledgements:

Citation:
Karyono G. Modified Wiener Filter using Scaled Median Absolute Deviation Normalized for Gaussian Noise Reduction. Computer Optics 2026; 50(3): 1724. doi: 10.18287/COJ1724.

References:

1. Abdulah CSK, Rahman R, Ahmad F, et al. Review study of image de-noising on digital image processing and applications. J Adv Res Appl Sci Eng Technol 2023; 30(1): 331-343. doi:10.37934/araset.30.1.331343.
2. Orazaev A, Lyakhov P, Baboshina V, Kalita D. Neural network system for recognizing images affected by random-valued impulse noise. Appl Sci 2023; 13(3): 1585. doi:10.3390/app13031585.
3. Hussain D, Gu YH. Exploring the impact of noise and image quality on deep learning performance in DXA images. Diagnostics 2024; 14(13): 1328. doi:10.3390/diagnostics14131328.
4. Honnatteppanavar J. Analyzing filtering techniques to eliminate Gaussian image noise. Natl Qual Res J 2021; 19(9): 1077-1084. doi:10.48047/nq.2021.19.9.NQ21182.
5. Jiang X, Xu S, Wu J, Zhou C, Ji S. Boosting noise reduction effect via unsupervised fine-tuning strategy. Appl Sci 2024; 14(5): 1742. doi:10.3390/app14051742.
6. Momeny M, Mohammad A, Agha M, Sheikhpour R. A noise-robust convolutional neural network for image classification. Results Eng 2021; 10: 100225. doi:10.1016/j.rineng.2021.100225.
7. Wang T, Hu Z, Guan Y. An efficient lightweight network for image denoising using progressive residual and convolutional attention feature fusion. Sci Rep 2024; 14: 12345. doi:10.1038/s41598-024-60139-x.
8. Hu J, Xia B, Li B, Yang W. Low-trace adaptation of zero-shot self-supervised blind image denoising. arXiv preprint 2024; arXiv:2403.12382. Available at: http://arxiv.org/abs/2403.12382.
9. Fan L, Zhang F, Fan H, Zhang C. Brief review of image denoising techniques. Vis Comput Ind Biomed Art 2019; 2(1): 7. doi:10.1186/s42492-019-0016-7.
10. Milanfar P, Delbracio M. Denoising: a powerful building-block for imaging, inverse problems, and machine learning. arXiv preprint 2024; arXiv:2409.06219. Available at: http://arxiv.org/abs/2409.06219.
11. Yadav M, Tiwari N. Performance comparison of image restoration techniques using CNN and their applications. In: Proc 5th Int Conf Comput Methodol Commun (ICCMC); 2021. pp. 1146-1151. doi:10.1109/ICCMC51019.2021.9418340.
12. Zhang K, Zuo W, Zhang L. FFDNet: toward a fast and flexible solution for CNN-based image denoising. IEEE Trans Image Process 2018; 27(9): 4608-4622. doi:10.1109/TIP.2018.2839891.
13. Anwar S, Barnes N. Real image denoising with feature attention. In: Proc IEEE Conf Comput Vis Pattern Recognit (CVPR); 2019. pp. 3155-3164. doi:10.1109/CVPR.2019.00327.
14. Kim Y, Soh JW, Park GY, Cho NI. Transfer learning from synthetic to real-noise denoising with adaptive instance normalization. In: Proc IEEE/CVF Conf Comput Vis Pattern Recognit (CVPR); 2020. pp. 3482-3492. doi:10.1109/CVPR42600.2020.00355.
15. Chen Y, Zhou G, Li L, Chen A, Wang Y, Li L. A high-quality self-supervised image denoising method based on SDDW-GAN and CHRNet. Expert Syst Appl 2024; 258: 125157. doi:10.1016/j.eswa.2024.125157.
16. Chen Y, Chen F. Application in insulator infrared image denoising with dual-tree complex wavelet transform. In: Proc 2020 12th Int Conf Meas Technol Mechatronics Autom (ICMTMA); 2020. pp. 249-252. doi:10.1109/ICMTMA50254.2020.00063.
17. Fan H, Zhang Y, Wang W, Li T. Suppressing seismic random noise based on non-subsampled shearlet transform and improved FFDNet. Front Earth Sci 2024; 12: 1408317. doi:10.3389/feart.2024.1408317.
18. Jiang N, Zhang Y, Li Q, Yan F. An infrared thermal image denoising method focusing on noise feature learning. Opt Laser Technol 2025; 184: 112475. doi:10.1016/j.optlastec.2025.112475.
19. Mohan R, Chokkalingam SP, Ganapathy K, Rama A. Comparative image quality analysis of spatial filters for pre-processing of CT abdominal images. Webology 2021; 18(Special Issue 4): 1449-1469. doi:10.14704/WEB/V18SI04/WEB18283.
20. Park CR, Kang SH, Lee Y. Median modified Wiener filter for improving the image quality of gamma camera images. Nucl Eng Technol 2020; 52(10): 2328-2333. doi:10.1016/j.net.2020.03.022.
21. Ju S, Kang SH, Lee Y. Optimization of mask size for median-modified Wiener filter according to matrix size of computed tomography images. Nucl Instrum Methods Phys Res A 2021; 1010: 165508. doi:10.1016/j.nima.2021.165508.
22. Lee S, Lee Y. Performance evaluation of median-modified Wiener filter algorithm in high-resolution CMOS radio-magnetic X-ray imaging system: an experimental study. Nucl Instrum Methods Phys Res A 2021; 1010: 165509. doi:10.1016/j.nima.2021.165509.
23. Zhou Z, Qu Y, Zhu B, Zhang B. Detection of typical transient signals in water by XGBoost classifier based on shape statistical features: application to the call of southern right whale. J Mar Sci Eng 2024; 12(9): 1596. doi:10.3390/jmse12091596.
24. Yan P, Jeong J, Sagan N, Weissman T. Universal discrete filtering with lookahead or delay. arXiv preprint 2025; arXiv:2501.10609. Available at: http://arxiv.org/abs/2501.10609.
25. Xiuyu Y, Liyong T, Feng C. Thermal infrared imaging for conveyor roller fault detection in coal mines. PLoS One 2024; 19(7): e0307591. doi:10.1371/journal.pone.0307591.
26. Howell DC. Median absolute deviation. In: Encyclopedia of statistics in behavioral science. Vol. 3. Chichester: Wiley; 2005. pp. 1193-1193. doi:10.1002/9781118445112.stat06232.

---

151, Molodogvardeiskaya str., Samara, 443001, Russia; E-mail: journal@computeroptics.ru; Tel: +7 (846) 242-41-24 (Executive secretary), +7 (846) 332-56-22 (Issuing editor), Fax: +7 (846) 332-56-20