Self-Supervised Generative Reconstruction Model for Blind Image Denoising and Detail Preservation in Adverse Visual Conditions

Paper ID: EIJTEM_2025_12_3_92-98

Author's Name: S Swarna Keerthi

Volume: 12

Issue: 3

Year: 2025

Page No: 92-98

Abstract:

Blind image denoising in adverse visual conditions remains one of the most challenging problems in low-level vision, particularly due to the absence of clean training targets and the complex, scene-dependent nature of real-world noise. This work presents a Self-Supervised Generative Reconstruction Model that restores clean image structures directly from noisy observations by leveraging intrinsic image priors, latent noise modeling, and generative reconstruction theory. The system consists of a noise encoder that learns hidden noise representations, a generative reconstruction network that projects the degraded input onto the natural image manifold, and a self-supervised noise-injection mechanism that enforces reconstruction consistency without requiring paired datasets. Experimental results across diverse scenes—including indoor, outdoor, and natural environments—demonstrate that the proposed framework effectively suppresses severe noise, recovers fine details, and preserves perceptual realism even under extreme degradations. Quantitative improvements in PSNR, SSIM, NIQE, LPIPS, and MSE validate the robustness and accuracy of the model, while qualitative comparisons show superior structural fidelity and visual clarity. The proposed approach provides a scalable and annotation-free solution for real-world denoising tasks, making it suitable for applications in surveillance, autonomous systems, mobile imaging, and scientific data restoration.

Keywords: Self-supervised denoising, generative reconstruction, image restoration, blind noise modelling, detail preservation, adverse visual conditions

References:

[1]. K. Dabov, A. Foi, V. Katkovnik, and K. Egiazarian, “Image denoising by sparse 3D transform-domain collaborative filtering,” IEEE Trans. Image Process., vol. 16, no. 8, pp. 2080–2095, Aug. 2007.
[2]. A. Buades, B. Coll, and J.-M. Morel, “A non-local algorithm for image denoising,” in Proc. CVPR, 2005, pp. 60–65.
[3]. P. Chatterjee and P. Milanfar, “Patch-based near-optimal image denoising,” IEEE Trans. Image Process., vol. 21, no. 4, pp. 1635–1649, Apr. 2012.
[4]. D. Tiwary and M. Javeed, “Power Management for Low Power VLSI Design,” International Journal of Creative Research Thoughts (IJCRT), vol. 6, no. 2, pp. 732–735, Apr. 2018.
https://www.ijcrt.org/viewfull.php?&p_id=IJCRT1892615
Google Scholar
[5]. K. Zhang, W. Zuo, Y. Chen, D. Meng, and L. Zhang, “Beyond a Gaussian denoiser: Residual learning of deep CNN for image denoising,” IEEE Trans. Image Process., vol. 26, no. 7, pp. 3142–3155, Jul. 2017.
[6]. X. Mao, C. Shen, and Y.–B. Yang, “Image restoration using very deep convolutional encoder–decoder networks with symmetric skip connections,” in Proc. NIPS, 2016, pp. 2802–2810.
[7]. C. Chen, Q. Chen, J. Xu, and V. Koltun, “Learning to see in the dark,” in Proc. CVPR, 2018, pp. 3291–3300.
[8]. J. Yue, T. Zhang, X. Zhou, and J. Yang, “Zero-reference deep curve estimation for low-light image enhancement,” in Proc. CVPR, 2020, pp. 1777–1786.
[9]. T. Mahesh, M. Javeed, and D. Tiwary, “Internet of Things for Health Care Mechanism,” International Journal of Research in Engineering, Science and Management, vol. 1, no. 9, pp. 1–6, Sep. 2018.
https://www.ijresm.com/Vol_1_2018/Vol1_Iss9_September18/IJRESM_19_1.pdf
Google Scholar
[10]. J. Batson and L. Royer, “Noise2Self: Blind denoising by self-supervision,” in Proc. ICML, 2019, pp. 524–533.
[11]. A. Krull, T.-O. Buchholz, and F. Jug, “Noise2Void—Learning denoising from single noisy images,” in Proc. CVPR, 2019, pp. 2129–2137.
[12]. Md. Javeed and G. Kande, “A Survey on Linking of Cell Tracks Using Segmentation,” International Journal of Applied Engineering Research, vol. 13, no. 7, pp. 5183–5189, 2018.
https://www.ripublication.com/ijaer18/ijaerv13n7_83.pdf
Google Scholar
[13]. J. Lehtinen et al., “Noise2Noise: Learning image restoration without clean data,” in Proc. ICML, 2018, pp. 2965–2974.
[14]. D. P. Kingma and M. Welling, “Auto-encoding variational Bayes,” in Proc. ICLR, 2014.
[15]. T. Srujana, D. Divya, and M. Javeed, “FPGA Implementation for Detection Disease on Cotton Plants Using Advanced Image Processing Algorithm,” International Journal of Advanced Research in Electrical, Electronics and Instrumentation Engineering, vol. 7, no. 11, pp. 4023–4030, Nov. 2018.
https://www.ijareeie.com/upload/2018/november/15_FPGA.pdf
Google Scholar
[16]. I. Goodfellow et al., “Generative adversarial nets,” in Proc. NIPS, 2014, pp. 2672–2680.
[17]. J. Ho, A. Jain, and P. Abbeel, “Denoising diffusion probabilistic models,” in Proc. NeurIPS, 2020, pp. 6840–6851.
[18]. Y. Song and S. Ermon, “Generative modeling by estimating gradients of the data distribution,” in Adv. Neural Inf. Process. Syst., 2019, pp. 11895–11907.
[19]. J. Zamir et al., “Restormer: Efficient Transformer for high-resolution image restoration,” in Proc. CVPR, 2022, pp. 5728–5739.
[20]. M. Senthil Kumar and M. Javeed, “Design of NoC Router with 3PE, Double and Triple Error Detection by Using Improved Hamming Code,” ARPN Journal of Engineering and Applied Sciences, vol. 14, no. 16, pp. 2874–2880, Aug. 2019.
https://www.arpnjournals.org/jeas/research_papers/rp_2019/jeas_0819_7884.pdf
Google Scholar
[21]. S. Guo, Z. Yan, K. Zhang, W. Zuo, and L. Zhang, “Toward convolutional blind denoising of real photographs,” in Proc. CVPR, 2019, pp. 1712–1722.
[22]. R. Timofte et al., “NTIRE 2020 challenge on real image denoising: Methods and results,” in Proc. CVPR Workshops, 2020, pp. 1–28.

View PDF