search for


Image restoration using convolutional denoising autoencoder in images
Journal of the Korean Data & Information Science Society 2020;31:25-40
Published online January 31, 2020;
© 2020 Korean Data and Information Science Society.

Jung Hun Song1 · Jeong Hee Kim2 · Dong Hoon Lim3

123Department of Information Statistics, Gyeongsang National University
Correspondence to: Professor, Department of Information Statistics and RINS, Gyeongsang National University, Jinju 52828, Korea. E-mail:
This research was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (Ministry of Science and ICT) (No. NRF-2017R1E1A1A03071057).
Received September 19, 2019; Revised October 14, 2019; Accepted October 24, 2019.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License ( which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
Digital images have been compromised by various noise in the process of acquisition, transmission and processing, resulting in the need for image restoration. Until now, methods of removing noise from images have used unique filters designed under certain distributions, which tend to be significantly less effective if the characteristics of the distribution are not met. In this paper, we are going to use the convolutional denoising autoencoder (CDAE) model of deep learning to eliminate noise. The CDAE model is a combination of the CNN (convolutional neural network) model and the DAE (denoising autoencoder) model, which is an applicable method regardless of the noise distribution of images. In order to evaluate the CDAE model proposed in this paper, we considered images damaged by various noises, Gaussian noise, impulse noise and speckle noise. We compared our CDAE model with CNN and traditional filters such as Mean filter, Median filter and Lee filter. Experimental results on several images show that the CDAE model yields significantly superior image quality and better PSNR (peak signalto- noise ratio) and MAE (mean absolute error).
Keywords : Convolutional denoising autoencoder (CDAE), deep learning, image restoration, noise reduction.