A Hybrid model of ResNet50 integrated U-Net for image Denoising for Polymer and Composite Microstructure Analysis

Notice

This is an unedited manuscript accepted for publication and provided as an Article in Press for early access at the author’s request. The article will undergo copyediting, typesetting, and galley proof review before final publication. Please be aware that errors may be identified during production that could affect the content. All legal disclaimers of the journal apply.

Year : 2026 | Volume : 14 | 03 | Page :
    By

    Akanksha Kochhar,

  • Rupali Pandey,

  • Aarti Sehwag,

  • Sanskar Kannaujia,

  • Anu Yadav,

  1. Assistant Professor, Department of Computer Science and Engineering, Bharati Vidyapeeth’s College of Engineering, New Delhi, India
  2. Assistant Professor, Department of Applied Sciences, Bharati Vidyapeeth’s College of Engineering, New Delhi, India
  3. Assistant Professor, Department of Computer Science and Engineering, Bharati Vidyapeeth’s College of Engineering, New Delhi, India
  4. Student, Department of Electronics and Communication Engineering, Bharati Vidyapeeth’s College of Engineering, New Delhi, India
  5. Assistant Professor, Department of Computer Science and Engineering, Bharati Vidyapeeth’s College of Engineering, New Delhi, India

Abstract

In digital era a high-quality imaging plays a very important role in polymer and composite material characterization features such as fiber-matrix interfaces, voids, microcracks cause problems in mechanical and functional properties. Polymer imaging includes optical microscopy and scanning electron microscopy, due to sensor limitation, environmental conditions add noise to the image and reduce quality of image. The noise degrades image quality, and it leads to reduce reliability in material analysis. So, there is need to reduce noise while preserving image features are desperately needed. To solve this problem, a hybrid model is introducing that uses deep learning models for noise reduction. Proposed method improves feature extraction by using skip connections and an attention mechanism, which prioritizes important information that is important for noise reduction. The method utilizes a diverse multimodal data set to train a robust noise reduction model based on the refined U-Net architecture, making it suitable for noise-prone imaging scenarios encountered in polymer science and materials engineering. Performance further enhanced with transfer learning from ResNet50. This approach successfully reduces noise. The proposed method shows strong capabilities to handle multi noise type and data quality across diverse applications, including polymer microstructure analysis, defect detection, and material quality assessment.

Keywords: Image Denoising, Skip Connections, U-Net, ResNet-50, Polymer Processing, Microstructure Inspection.

How to cite this article:
Akanksha Kochhar, Rupali Pandey, Aarti Sehwag, Sanskar Kannaujia, Anu Yadav. A Hybrid model of ResNet50 integrated U-Net for image Denoising for Polymer and Composite Microstructure Analysis. Journal of Polymer & Composites. 2026; 14(03):-.
How to cite this URL:
Akanksha Kochhar, Rupali Pandey, Aarti Sehwag, Sanskar Kannaujia, Anu Yadav. A Hybrid model of ResNet50 integrated U-Net for image Denoising for Polymer and Composite Microstructure Analysis. Journal of Polymer & Composites. 2026; 14(03):-. Available from: https://journals.stmjournals.com/jopc/article=2026/view=243545


References

[1]      E. ÇETİNKAYA, M. F. KIRAÇ, E. ÇETİNKAYA, and M. F. KIRAÇ, “Image denoising using deep convolutional autoencoder with feature pyramids,” Turkish Journal of Electrical Engineering and Computer Sciences, vol. 28, no. 4, pp. 2096–2109, Jan. 2020, doi: 10.3906/elk-1911-138.

[2]      T. Salimans, I. Goodfellow, W. Zaremba, V. Cheung, A. Radford, and X. Chen, “Improved Techniques for Training GANs,” Adv. Neural Inf. Process. Syst., pp. 2234–2242, Jun. 2016, Accessed: May 10, 2026. [Online]. Available: https://arxiv.org/pdf/1606.03498

[3]      L. Gonog and Y. Zhou, “A review: Generative adversarial networks,” Proceedings of the 14th IEEE Conference on Industrial Electronics and Applications, ICIEA 2019, pp. 505–510, Jun. 2019, doi: 10.1109/ICIEA.2019.8833686.

[4]      O. Kupyn, V. Budzan, M. Mykhailych, D. Mishkin, and J. Matas, “DeblurGAN: Blind Motion Deblurring Using Conditional Adversarial Networks”, Accessed: May 10, 2026. [Online]. Available: https://github.com/KupynOrest/DeblurGAN

[5]      C. Ledig et al., “Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network,” Proceedings – 30th IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017, vol. 2017-January, pp. 105–114, Sep. 2016, doi: 10.1109/CVPR.2017.19.

[6]      J. Zhu, J. Zhang, Y. Cao, and Z. Wang, “Image Guided Depth Enhancement via Deep Fusion and Local Linear Regularization”.

[7]      P. Riot, A. Almansa, Y. Gousseau, F. Tupin, and A. Almansa, “Penalizing local correlations in the residual improves image denoising performance”, Accessed: May 10, 2026. [Online]. Available: https://hal.science/hal-01341968v1

[8]      R. Li, C.-T. Li, and Y. Guan, “INCREMENTAL UPDATE OF FEATURE EXTRACTOR FOR CAMERA IDENTIFICATION”.

[9]      Y. Gan, E. Angelini, A. Laine, and C. Hendon, “BM3D-based ultrasound image denoising via brushlet thresholding,” Proceedings – International Symposium on Biomedical Imaging, vol. 2015-July, pp. 667–670, Jul. 2015, doi: 10.1109/ISBI.2015.7163961.

[10]    A. Scholefield and P. L. Dragotti, “IEEE TRANSACTIONS ON IMAGE PROCESSING 1 Quadtree Structured Image Approximation for Denoising and Interpolation,” 2013.

[11]    J. Dong, A. Kandemir, and I. Hamerton, “Microstructural characterisation of fibre-hybrid polymer composites using U-Net on optical images,” Compos. Part A Appl. Sci. Manuf., vol. 190, p. 108569, Mar. 2025, doi: 10.1016/J.COMPOSITESA.2024.108569.

[12]    K. Thomas, “International Journal of Engineering & Advanced Technology (IJEAT),” International Journal of Engineering and Advanced Technology (IJEAT), pp. 2249–8958, 2020, doi: 10.35940/ijeat.F1555.1010120.

[13]    W. El-Shafai et al., “Efficient Deep-Learning-Based Autoencoder Denoising Approach for Medical Image Diagnosis,” Computers, Materials & Continua, vol. 70, no. 3, pp. 6107–6125, Oct. 2021, doi: 10.32604/CMC.2022.020698.

[14]    P. Venkataraman, “Image Denoising Using Convolutional Autoencoder”.

[15]    A. P. Patii, A. Pramod, A. Harish, K. Singh, and K. Purushotham, “An approach to image denoising using autoencoders and spatial filters for Gaussian noise,” Proceedings of the Confluence 2021: 11th International Conference on Cloud Computing, Data Science and Engineering, pp. 454–458, Jan. 2021, doi: 10.1109/CONFLUENCE51648.2021.9377166.

[16]    M. K. , N. R. , K. E. H. , A. H. E. and O. M. Jasim, “ Image noise removal techniques: A comparative analysis,” International Journal of Science and Applied Information Technology, pp. 24–29, 2019.

[17]    C. Tian, Y. Xu, L. Fei, and K. Yan, “Deep learning for image denoising: A survey,” SpringerC Tian, Y Xu, L Fei, K YanInternational Conference on Genetic and Evolutionary Computing, 2018•Springer, vol. 834, pp. 563–572, 2019, doi: 10.1007/978-981-13-5841-8_59.

[18]    A. Raj, “Image Denoising Using Python and Machine Learning,” Int. J. Res. Appl. Sci. Eng. Technol., vol. 11, no. 5, pp. 7230–7235, May 2023, doi: 10.22214/IJRASET.2023.53406.

[19]    Z. Wang, L. Wang, S. Duan, and Y. Li, “An Image Denoising Method Based on Deep Residual GAN,” J. Phys. Conf. Ser., vol. 1550, no. 3, Jun. 2020, doi: 10.1088/1742-6596/1550/3/032127.

[20]    C. Ruikai, “Research Progress in Image Denoising Algorithms Based on Deep Learning,” J. Phys. Conf. Ser., vol. 1345, no. 4, Nov. 2019, doi: 10.1088/1742-6596/1345/4/042055.

[21]    L. Shen, B. Zhao, Q. Li, C. Zhang, X. Sun, and B. Peng, “Local to non-local: Multi-scale progressive attention network for image restoration,” Computer Vision and Image Understanding, vol. 233, Aug. 2023, doi: 10.1016/j.cviu.2023.103725.

[22]    M. Zubair, H. Md Rais, and T. Alazemi, “A Novel Attention-Guided Enhanced U-Net With Hybrid Edge-Preserving Structural Loss for Low-Dose CT Image Denoising,” IEEE Access, vol. 13, pp. 6909–6923, Jan. 2025, doi: 10.1109/ACCESS.2025.3526619.

[23]    L. Gonog and Y. Zhou, “A review: Generative adversarial networks,” Proceedings of the 14th IEEE Conference on Industrial Electronics and Applications, ICIEA 2019, pp. 505–510, Jun. 2019, doi: 10.1109/ICIEA.2019.8833686.

[24]    Z. Zhang, X. Du, F. Gao, L. Chen, and B. Li, “An improved u-net method for denoising ultrasonic echo signals in carbon fiber composites,” Ultrasonics, vol. 157, p. 107782, Jan. 2026, doi: 10.1016/J.ULTRAS.2025.107782).

Ahead of Print Subscription Review Article
Volume 14
03
Received 29/01/2026
Accepted 14/03/2026
Published 12/05/2026
Publication Time 103 Days
1

Login


My IP

PlumX Metrics