Wasserstein generative adversarial network based De-Blurring using perceptual similarity

Minsoo Hong, Yoonsik Choe

Research output: Contribution to journalArticle

Abstract

The de-blurring of blurred images is one of the most important image processing methods and it can be used for the preprocessing step in many multimedia and computer vision applications. Recently, de-blurring methods have been performed by neural network methods, such as the generative adversarial network (GAN), which is a powerful generative network. Among many different types of GAN, the proposed method is performed using theWasserstein generative adversarial network with gradient penalty (WGANGP). Since edge information is the most important factor in an image, the style loss function is applied to represent the perceptual information of the edge in order to preserve small edge information and capture its perceptual similarity. As a result, the proposed method improves the similarity between sharp and blurred images by minimizing the Wasserstein distance, and it captures well the perceptual similarity using the style loss function, considering the correlation of features in the convolutional neural network (CNN). To confirm the performance of the proposed method, three experiments are conducted using two datasets: the GOPRO Large and Kohler dataset. The optimal solutions are found by changing the parameter values experimentally. Consequently, the experiments depict that the proposed method achieves 0.98 higher performance in structural similarity (SSIM) and outperforms other de-blurring methods in the case of both datasets.

Original languageEnglish
Article number2358
JournalApplied Sciences (Switzerland)
Volume9
Issue number11
DOIs
Publication statusPublished - 2019 Jun 1

Fingerprint

blurring
Neural networks
Computer vision
Image processing
Experiments
computer vision
multimedia
preprocessing
penalties
image processing
gradients

All Science Journal Classification (ASJC) codes

  • Materials Science(all)
  • Instrumentation
  • Engineering(all)
  • Process Chemistry and Technology
  • Computer Science Applications
  • Fluid Flow and Transfer Processes

Cite this

@article{cac914292f8c46ceba895951a1ebee57,
title = "Wasserstein generative adversarial network based De-Blurring using perceptual similarity",
abstract = "The de-blurring of blurred images is one of the most important image processing methods and it can be used for the preprocessing step in many multimedia and computer vision applications. Recently, de-blurring methods have been performed by neural network methods, such as the generative adversarial network (GAN), which is a powerful generative network. Among many different types of GAN, the proposed method is performed using theWasserstein generative adversarial network with gradient penalty (WGANGP). Since edge information is the most important factor in an image, the style loss function is applied to represent the perceptual information of the edge in order to preserve small edge information and capture its perceptual similarity. As a result, the proposed method improves the similarity between sharp and blurred images by minimizing the Wasserstein distance, and it captures well the perceptual similarity using the style loss function, considering the correlation of features in the convolutional neural network (CNN). To confirm the performance of the proposed method, three experiments are conducted using two datasets: the GOPRO Large and Kohler dataset. The optimal solutions are found by changing the parameter values experimentally. Consequently, the experiments depict that the proposed method achieves 0.98 higher performance in structural similarity (SSIM) and outperforms other de-blurring methods in the case of both datasets.",
author = "Minsoo Hong and Yoonsik Choe",
year = "2019",
month = "6",
day = "1",
doi = "10.3390/app9112358",
language = "English",
volume = "9",
journal = "Applied Sciences (Switzerland)",
issn = "2076-3417",
publisher = "Multidisciplinary Digital Publishing Institute (MDPI)",
number = "11",

}

Wasserstein generative adversarial network based De-Blurring using perceptual similarity. / Hong, Minsoo; Choe, Yoonsik.

In: Applied Sciences (Switzerland), Vol. 9, No. 11, 2358, 01.06.2019.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Wasserstein generative adversarial network based De-Blurring using perceptual similarity

AU - Hong, Minsoo

AU - Choe, Yoonsik

PY - 2019/6/1

Y1 - 2019/6/1

N2 - The de-blurring of blurred images is one of the most important image processing methods and it can be used for the preprocessing step in many multimedia and computer vision applications. Recently, de-blurring methods have been performed by neural network methods, such as the generative adversarial network (GAN), which is a powerful generative network. Among many different types of GAN, the proposed method is performed using theWasserstein generative adversarial network with gradient penalty (WGANGP). Since edge information is the most important factor in an image, the style loss function is applied to represent the perceptual information of the edge in order to preserve small edge information and capture its perceptual similarity. As a result, the proposed method improves the similarity between sharp and blurred images by minimizing the Wasserstein distance, and it captures well the perceptual similarity using the style loss function, considering the correlation of features in the convolutional neural network (CNN). To confirm the performance of the proposed method, three experiments are conducted using two datasets: the GOPRO Large and Kohler dataset. The optimal solutions are found by changing the parameter values experimentally. Consequently, the experiments depict that the proposed method achieves 0.98 higher performance in structural similarity (SSIM) and outperforms other de-blurring methods in the case of both datasets.

AB - The de-blurring of blurred images is one of the most important image processing methods and it can be used for the preprocessing step in many multimedia and computer vision applications. Recently, de-blurring methods have been performed by neural network methods, such as the generative adversarial network (GAN), which is a powerful generative network. Among many different types of GAN, the proposed method is performed using theWasserstein generative adversarial network with gradient penalty (WGANGP). Since edge information is the most important factor in an image, the style loss function is applied to represent the perceptual information of the edge in order to preserve small edge information and capture its perceptual similarity. As a result, the proposed method improves the similarity between sharp and blurred images by minimizing the Wasserstein distance, and it captures well the perceptual similarity using the style loss function, considering the correlation of features in the convolutional neural network (CNN). To confirm the performance of the proposed method, three experiments are conducted using two datasets: the GOPRO Large and Kohler dataset. The optimal solutions are found by changing the parameter values experimentally. Consequently, the experiments depict that the proposed method achieves 0.98 higher performance in structural similarity (SSIM) and outperforms other de-blurring methods in the case of both datasets.

UR - http://www.scopus.com/inward/record.url?scp=85067263598&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85067263598&partnerID=8YFLogxK

U2 - 10.3390/app9112358

DO - 10.3390/app9112358

M3 - Article

AN - SCOPUS:85067263598

VL - 9

JO - Applied Sciences (Switzerland)

JF - Applied Sciences (Switzerland)

SN - 2076-3417

IS - 11

M1 - 2358

ER -