Skip to main content
SpringerLink
Log in

Optimal image Denoising using patch-based convolutional neural network architecture

  • Published:
Multimedia Tools and Applications Aims and scope Submit manuscript

Abstract

Traditional CNN uses fixed location which is irrelevant and they show the inability to capture edges and texture which causes the smoothness of artefacts, thus many details are lost. Hence, this research work designs and develops a novel CNN-backed architecture i.e. PCNN (Patch-based CNN). PCNN sets network depth based on patch size. Moreover, research work follows various steps, first patch similarity identification is carried out later it is given to the designed customized CNN for denoising, and then these patches are integrated to achieve the efficient denoised image. Performance evaluation of the proposed PCNN architecture was carried out on the KODAK dataset by comparing and considering different parameters like PSNR and SSIM. A further different signal level is used for deep evaluation. Comparative analysis with different state-of-art techniques including deep learning-based shows that PCNN simply outperforms the other model. Further analysis is carried out on three random images i.e. image 8, image 19, and image 20 are considered for PSNR and SSIM comparison and it is observed that in terms of PSNR, PCNN achieves 93.03%, 62.01% and 67.78% improvised over the existing model. Further, SSIM is considered for the same images and PCNN achieves 42.55%, 19.84%, and 15.63% improvisation over the existing model. Further, this research compares PSNR values over different signal levels of 2, 4, 6, 8, and 10 and achieves improvisation of 45.51%, 38.72%, 38.57%, 56.2%, and 35.06% respectively. Furthermore, Average PSNR is compared considering Red, Green, Blue, and RGB, PCNN achieves improvisation of 5.65%, 2.34%, 8.62%, and 4.87%.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7

Similar content being viewed by others

Data availability

The datasets generated during and/or analysed during the current study are available in the reference [22].

References

  1. Afonso MV, Bioucas-Dias JM, Figueiredo MAT (2010) An augmented lagrangian approach to the constrained optimization formulation of imaging inverse problems, Image Processing, IEEE Transactions on. PP(99, 1)

  2. Buades A, Coll B, Morel J-M (2005) A non-local algorithm for image denoising. In: Proc. of Computer Vision and Pattern Recognition (CVPR). pp. 60–65

  3. Burger HC, Schuler CJ, Harmeling S (2012) Image denoising: Can plain neural networks compete with bm3d?” in Proc. of Computer Vision and Pattern Recognition (CVPR). pp. 2392–2399

  4. Condat L, Mosaddegh S “Joint demosaicking and denoising by total variation minimization,” in Proc. 19th IEEE Int. Conf. Image Process., Sep. 2012, pp. 2781–2784.

  5. Cui K, Jin Z, Steinbach E (2018) Color image demosaicking using a 3-stage convolutional neural network structure. In: Proc. 25th IEEE Int. Conf. Image Process. pp. 2177–2181

  6. Cui K, Boev A, Alshina E, Steinbach E (2021) Color image restoration exploiting Inter-Channel correlation with a 3-stage CNN. IEEE J Sel Top Signal Proc 15(2):174–189. https://doi.org/10.1109/JSTSP.2020.3043148

    Article  Google Scholar 

  7. Danielyan A, Katkovnik V, Egiazarian K (2010) Image deblurring by augmented lagrangian with bm3d frame prior. In: Workshop on Information Theoretic Methods in Science and Engineering, WITMSE 2010, Tampere, Finland

  8. Dong C, Loy CC, He K, Tang X (2014) Learning a deep convolutional network for image super-resolution. In: Proc. of European Conference on Computer Vision (ECCV), 2014. pp. 184–199

  9. Dong N, Maggioni M, Yang Y, P'erez-Pellitero E, Leonardis A, McDonagh SG (2021) Residual contrastive learning for joint Demosaicking and Denoising. ArXiv, abs/2106.10070

  10. Elad M (2010) Sparse and redundant representations: from theory to applications in signal and image processing. Springer Press

    Book  MATH  Google Scholar 

  11. Elgendy OA, Gnanasambandam A, Chan SH, Ma J (2021) Low-light Demosaicking and Denoising for small pixels using learned frequency selection. IEEE Trans Comput Imag 7:137–150. https://doi.org/10.1109/TCI.2021.3052694

    Article  Google Scholar 

  12. Gharbi M, Chaurasia G, Paris S, Durand F (2016) Deep joint demosaicking and denoising. ACM Trans Graph 35:191:1–191:12

    Article  Google Scholar 

  13. Gnanasambandam A, Elgendy O, Ma J, Chan SH (2019) Megapixel photon-counting colour imaging using a quanta image sensor. Opt Express 27:17298–17310

    Article  Google Scholar 

  14. Gu S, Zhang L, Zuo W, Feng X (2014) Weighted nuclear norm minimization with application to image denoising. In: Proc. of Computer Vision and Pattern Recognition (CVPR). pp. 2862–2869

  15. Guo S, Liang Z, Zhang L (2021) Joint Denoising and DemosaickingWith Green Channel prior for real-world burst images. IEEE Trans Image Process 30:6930–6942. https://doi.org/10.1109/TIP.2021.3100312

    Article  MathSciNet  Google Scholar 

  16. Hirakawa K, Parks TW (2005) Adaptive homogeneity-directed demosaicing algorithm. IEEE Trans Image Process 14(3):360–369

    Article  Google Scholar 

  17. Ioffe S, Szegedy C (2015) Batch normalization: Accelerating deep network training by reducing internal covariate shift. In: Proc. of International Conference on Machine Learning (ICML). pp. 448–456

  18. Jain V, Seung S (2009) Natural image denoising with convolutional networks,” in Advances in Neural Information Processing Systems (NIPS). pp. 769–776

  19. Khadidos AO, Khadidos AO, Khan FQ, Tsaramirsis G, Ahmad A (2021) Bayer image demosaicking and denoising based on specialized networks using deep learning. Multimedia Systems 27:807–819. https://doi.org/10.1007/s00530-020-00707-z

    Article  Google Scholar 

  20. Kiku D, Monno Y, Tanaka M, Okutomi M (2014) Minimized-laplacian residual interpolation for colour image demosaicking. I+n: Proc. SPIE 9023, Digit. Photogr. X, vol. 9023. Art no 90230L

  21. Kiku D, Monno Y, Tanaka M, Okutomi M (2016) Beyond colour difference: residual interpolation for colour image demosaicking. IEEE Trans Image Process 25(3):1288–1300

    Article  MathSciNet  MATH  Google Scholar 

  22. “Kodak lossless true color image suite,” (n.d.) [Online]. Available: http://r0k.us/graphics/kodak/

  23. Kokkinos F, Lefkimmiatis S (2019) Iterative joint image demosaicking and denoising using a residual denoising network. IEEE Trans Image Process 28(8):4177–4188

    Article  MathSciNet  MATH  Google Scholar 

  24. Lee M, Kim H, Paik J (2019) Correction of Barrel Distortion in Fisheye Lens Images Using Image-Based Estimation of Distortion Parameters IEEE Access, 11. https://doi.org/10.1109/access.2019.290841

  25. Li Q, Zhu Z, Xu C, Tang Y (2017) A novel denoising method for acoustic signal. 2017 IEEE International Conference on Signal Processing, Communications and Computing (SPCC), 2017. pp. 1–5. https://doi.org/10.1109/ICSPCC.2017.8242453

  26. Liang Z, Cai J, Cao Z, Zhang L (2021) Camera net: a two-stage framework for effective camera ISP learning. IEEE Trans Image Process 30:2248–2262. https://doi.org/10.1109/tip.2021.3051486

    Article  Google Scholar 

  27. Liu D, Wen B, Liu X, Huang TS (2017) When image denoising meets high-level vision tasks: a deep learning approach. CoRR

  28. Monno Y, Kiku D, Tanaka M, Okutomi M (2017) Adaptive residual interpolation for colour and multispectral image demosaicking. Sensors 17(12)

  29. Philbin J, Chum O, Isard M, Sivic J, Zisserman A (2007) Object retrieval with large vocabularies and fast spatial matching. In: Proceedings of the IEEE conference on computer vision and pattern recognition

  30. Shao K, Fan Q, Zhang Y, Bao F, Zhang C (2021) Noisy single image super-resolution based on local fractal feature analysis. IEEE Access 9:33385–33395. https://doi.org/10.1109/ACCESS.2021.3061118

    Article  Google Scholar 

  31. Shao K, Fan Q, Zhang Y, Bao F, Zhang C (2021) Noisy single image super-resolution based on local fractal feature analysis. IEEE Access 9:33385–33395. https://doi.org/10.1109/ACCESS.2021.3061118

    Article  Google Scholar 

  32. Shi B, Lian Q, Chen S, Fan X (2018) Sbm3d: sparse regularization model induced by bm3d for weighted diffraction imaging. IEEE Access 6:46266–46280

    Article  Google Scholar 

  33. R. Tan, K. Zhang, W. Zuo, and L. Zhang, “Color image demosaicking via deep residual learning,” in Proc. IEEE Int. Conf. Multimedia Expo., Jul. 2017, pp. 793–798.

  34. Yang D, Sun J (2018) Bm3d-net: a convolutional neural network for transform-domain collaborative filtering. IEEE Signal Proc Lett 25(1):55–59

    Article  Google Scholar 

  35. Zhang L, Wu X (2005) Color demosaicking via directional linear minimum mean square error estimation. IEEE Trans Image Process 14(12):2167–2178

    Article  Google Scholar 

  36. Zhang L, Zuo W (2017) Image restoration: from sparse and low-rank priors to deep priors [lecture notes]. IEEE Signal Process Mag 34(5):172–179

    Article  Google Scholar 

  37. Zhang L, Wu X, Buades A, Li X (2011) Color demosaicking by local directional interpolation and nonlocal adaptive thresholding. J Electron Imag 20(2):1–17

    Google Scholar 

  38. Zhang K, Zuo W, Chen Y, Meng D, Zhang L (2017) Beyond a Gaussian denoiser: residual learning of deep CNN for image denoising. IEEE Trans Image Process 26(7):3142–3155

    Article  MathSciNet  MATH  Google Scholar 

  39. Zhang K, Zuo W, Chen Y, Meng D, Zhang L (2017) Beyond a Gaussian denoiser: residual learning of deep CNN for image denoising. IEEE Trans Image Process 26(7):3142–3155

    Article  MathSciNet  MATH  Google Scholar 

Download references

Author information

Authors and Affiliations

  1. K.B.N college of engineering, Kalaburagi, Karnataka, India

    Shabana Tabassum

  2. Bheemanna Khandre Institute of Technology, Bhalki, Karnataka, India

    SanjayKumar C Gowre

Authors
  1. Shabana Tabassum
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. SanjayKumar C Gowre
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Shabana Tabassum.

Ethics declarations

Conflict of interest

The authors declare no conflict of interest.

Additional information

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Tabassum, S., Gowre, S.C. Optimal image Denoising using patch-based convolutional neural network architecture. Multimed Tools Appl 82, 29805–29821 (2023). https://doi.org/10.1007/s11042-023-15014-8

Download citation

  • Received:

  • Revised:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s11042-023-15014-8

Keywords

Search

Navigation