Abstract
Improving the quality of a noisy image is important for image applications. Many novel schemes pay great efforts in the removal of impulse noise. Most of them restore noisy pixels only by using the neighboring noise-free pixels, but the relationship between a noisy image and its noise-free one, which denotes the clean image not corrupted by noise, is ignored. So the reconstruction quality cannot be further improved. In this study, we employ a deep-learning fully connected neural network (FCNN) to select top N candidates of neighboring un-corrupted pixels for the restoration of a center noisy pixel in an analysis window. Hence, the mean value of the gray levels of these top N pixels is computed and employed to replace the noisy pixel, yielding the noisy pixel being restored. The experimental results reveal that the proposed deep-learning FCNN mean filter can remove impulse noise effectively in corrupted images with different noise densities.
Similar content being viewed by others
Abbreviations
- \(\varPsi_{i,j}\) :
-
Feature vector of the FCNN
- \(\eta_{i,j}\) :
-
Noise-free flag
- \(r_{i,j}\) :
-
Spatial distance
- \(\, k_{i,j}\) :
-
Similarity measure
- \(\, \sigma_{i,j}\) :
-
Pixel deviation
- \(x_{i,j}\) :
-
Noisy pixel
- \(d_{i,j}\) :
-
The gray-level distance
- \(N_{\text{ne}}\) :
-
The number of non-extreme pixel
- \(\, \tilde{x}_{i,j}\) :
-
Non-corrupted pixel
- \(l_{i,j}^{*}\) :
-
The optimum position index of the neighboring noise-free pixel
- \(s_{i,j}\) :
-
The pixel in the noise-free image
- \(\hat{l}_{i,j}\) :
-
The recognized index of neighboring pixel
- \(\tilde{s}_{i,j}\) :
-
The recognized pixel of the FCNN in an analysis window
- \(\hat{s}_{i,j}\) :
-
Denoised pixel
- \(\tilde{x}_{\text{Top}\,i}\) :
-
The top i recognized results of the FCNN
- N :
-
The number of noise-free pixels among top i recognized results
- \(N_{s}\) :
-
The desired number of neighboring noise-free pixels recognized by the FCNN in an analysis window
- \(F_{\text{Top}\,i}\) :
-
The noise-free flag of the top ith neighboring pixel
- \(s_{\text{max} }\) :
-
The maximum value of the gray level
- \(\varepsilon\) :
-
The mean square error (MSE) between the noise-free image and the enhanced one
- Q :
-
The number of columns in an image
- P :
-
The number of rows in an image
- \(\mu_{{s_{\text{m}} }}\) :
-
The mean of the noise-free pixels
- \(\mu_{{\hat{s}_{\text{m}} }}\) :
-
The mean of the reconstructed pixels
- \(\sigma_{{s_{\text{m}} }}\) :
-
The standard deviations of clean pixels
- \(\sigma_{{\hat{s}_{\text{m}} }}\) :
-
The standard deviations of reconstructed pixels
- \(\sigma_{{s_{\text{m}} \hat{s}_{\text{m}} }}\) :
-
The square root of the covariance of clean and reconstructed pixels
- MSSIM:
-
Mean structural similarity index
References
Dong Y, Xu S (2007) A new directional weighted median filter for removal of random-valued impulse noise. IEEE Signal Process Lett 14:193–196. https://doi.org/10.1109/LSP.2006.884014
Lu CT, Chou CT (2012) Denoising of salt-and-pepper noise corrupted image using modified directional-weighted-median filter. Pattern Recognit Lett 33:1287–1295. https://doi.org/10.1016/j.patrec.2012.03.025
Ng PE, Ma KK (2006) A switching median filter with boundary discriminative noise detection for extremely corrupted images. IEEE Trans Image Process 15:1506–1516. https://doi.org/10.1109/TIP.2005.871129
Esakkirajan S, Veerakumar T, Subramanyam AN, PremChand CH (2011) Removal of high density salt and pepper noise through modified decision based unsymmetric trimmed median filter. IEEE Signal Process Lett 18:287–290. https://doi.org/10.1109/LSP.2011.2122333
Li Z, Liu G, Xu Y, Cheng Y (2014) Modified directional weighted filter for removal of salt & pepper noise. Pattern Recogit Lett 40:113–120. https://doi.org/10.1016/j.patrec.2013.12.022
Zhang X, Xiong Y (2009) Impulse noise removal using directional difference based noise detector and adaptive weighted mean filter. IEEE Signal Process Lett 16:295–298. https://doi.org/10.1109/LSP.2009.2014293
Toh KKV, Mat-Isa NA (2010) Noise adaptive fuzzy switching median filter for salt-and-pepper noise reduction. IEEE Signal Process Lett 17:281–284. https://doi.org/10.1109/LSP.2009.2038769
Lu CT, Chen YY, Wang LL, Chang CF (2016) Removal of salt-and- pepper noise in corrupted image using three-values-weighted approach with variable-size window. Pattern Recognit Lett 80:188–199. https://doi.org/10.1016/j.patrec.2016.06.026
Lu CT, Chen MY, Shen JH, Wang LL, Hsu CC (2018) Removal of salt-and-pepper noise for X-ray bio-images using pixel-variation gain factors. Comput Electr Eng 71:862–876. https://doi.org/10.1016/j.compeleceng.2017.08.012
Lee JY, Jung SY, Kim PW (2018) Adaptive switching filter for impulse noise removal in digital content. Soft Comput 22:1445–1455. https://doi.org/10.1007/s00500-017-2843-9
Schuster T, Sussner P (2017) An adaptive image filter based on the fuzzy transform for impulse noise reduction. Soft Comput 21:3659–3672. https://doi.org/10.1007/s00500-017-2669-5
Latorre-Carmona P, Miñana J, Morillas S (2020) Colour image denoising by eigenvector analysis of neighbourhood colour samples. Signal Image Video Process 14:483–490. https://doi.org/10.1007/s11760-019-01575-5
Lu CT, Shen JH, Wang LL, Hsu CC, Liu LL (2018) Impulse noise denoising using confidence measure with non-sequential process order for X-ray Bio-images. J Med Biol Eng 38:905–916. https://doi.org/10.1007/s40846-017-0356-8
Erkan U, Thanh DNH, Hieu LM, Engínoğlu S (2019) An iterative mean filter for image denoising. IEEE Access 7:167847–167859. https://doi.org/10.1109/ACCESS.2019.2953924
Yang Q, Yan P, Zhang Y, Yu H, Shi Y, Mou X, Kalra MK, Zhang Y, Sun L, Wang G (2018) Low-dose CT image denoising using a generative adversarial network with Wasserstein distance and perceptual loss. IEEE Trans Med Imaging 37:1348–1357. https://doi.org/10.1109/TMI.2018.2827462
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 26:3142–3155. https://doi.org/10.1109/TIP.2017.2662206
Zhang K, Zuo W, Zhang L (2018) FFDNet: toward a fast and flexible solution for CNN-based image denoising. IEEE Trans Image Process 27:4608–4622. https://doi.org/10.1109/TIP.2018.2839891
Chen RH, Lu CT, Wang LL, Lin CA, Shen JH (2019) Removal of salt-and-pepper noise using convolutional-neural networks. In: Proceedings of Conference on Information Technology Applications Outlying Islands, pp 590–597
Niu Y, Yang Y, Guo W, Lin L (2018) Region-aware image denoising by exploring parameter preference. IEEE Trans Circuits Syst. Video Technol 28:2433–2438. https://doi.org/10.1109/TCSVT.2018.2859982
Liu Z, Ya WQ, Yang ML (2018) Image denoising based on a CNN model. In: Proceedings of International Conference on Control Automation Robotics, pp 389–393
Isogawa K, Shiodera T, Takeguchi T (2018) Deep shrinkage convolutional neural network for adaptive noise reduction. IEEE Signal Process Lett 25:224–228. https://doi.org/10.1109/LSP.2017.2782270
Lan R, Zou H, Pang C, Zhong Y, Liu Z, Luo X (2019) Image denoising via deep residual convolutional neural networks. Signal Image Video Process. https://doi.org/10.1007/s11760-019-01537-x
Jifara W, Jiang F, Rho S, Cheng M, Liu S (2019) Medical image denoising using convolutional neural network: a residual learning approach. J Supercomput 75:704–718. https://doi.org/10.1007/s11227-017-2080-0
Sangaiah AK, Medhane DV, Han T, Hossain MS, Muhammad G (2019) Enforcing position-based confidentiality with machine learning paradigm through mobile edge computing in real-time industrial informatics. IEEE Trans Ind Inf 15:4189–4196. https://doi.org/10.1109/tii.2019.2898174
Sangaiah AK, Medhane DV, Bian G, Ghoneim A, Alrashoud M, Hossain MS (2020) Energy-aware green adversary model for cyberphysical security in industrial system. IEEE Trans Ind Inf 16:3322–3329. https://doi.org/10.1109/TII.2019.2953289
Xing Y, Xu J, Tan J, Li D, Zha W (2019) Deep CNN for removal of salt and pepper noise. IET Image Process 13:1550–1560. https://doi.org/10.1049/iet-ipr.2018.6004
Li Z, Cheng Y, Tang K, Xu Y, Zhang D (2015) A salt & pepper noise filter based on local and global image information. Neurocomput 159:172–185. https://doi.org/10.1016/j.neucom.2014.12.087
Liu J, Liu R, Wang Y, Chen YJ, Yang Y, Ma D (2017) Image denoising searching similar blocks along edge directions. Signal Process Image Commun 57:33–45. https://doi.org/10.1016/j.image.2017.05.001
Zhang M, Desrosiers C (2018) Structure preserving image denoising based on low-rank reconstruction and gradient histograms. Comput Vis Image Underst 171:48–60. https://doi.org/10.1016/j.cviu.2018.05.006
Khmag A, Ramli AR, Kamarudin N (2019) Clustering-based natural image denoising using dictionary learning approach in wavelet domain. Soft Comput 23:8013–8027. https://doi.org/10.1007/s00500-018-3438-9
Guo Q, Zhang C, Zhang Y, Liu H (2016) An efficient SVD-based method for image denoising. IEEE Trans Circuits Syst Video Technol 26:868–880. https://doi.org/10.1109/TCSVT.2015.2416631
Fan L, Li X, Fan H, Feng Y, Zhang C (2019) Adaptive texture-preserving denoising method using gradient histogram and nonlocal self-similarity priors. IEEE Trans Circuits Syst Video Technol 29:3222–3235. https://doi.org/10.1109/TCSVT.2018.2878794
Wang Z, Bovik AC, Sheikh HR, Simoncelli EP (2004) Image quality assessment: from error visibility to structural similarity. IEEE Trans Image Process 13:600–612. https://doi.org/10.1109/TIP.2003.819861
Acknowledgements
This work was supported by the Ministry of Science and Technology, Taiwan [Grant Number MOST 104-2221-E-468-007]. Our gratitude goes to the reviewers for their valuable comments which have improved the quality of this paper.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Lu, CT., Wang, LL., Shen, JH. et al. Image enhancement using deep-learning fully connected neural network mean filter. J Supercomput 77, 3144–3164 (2021). https://doi.org/10.1007/s11227-020-03389-6
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11227-020-03389-6