One-to-many Image super resolution algorithm based on generative adversarial networks

Liu Lin; Jiang Yu

doi:10.1117/12.2673371

3 April 2023 One-to-many Image super resolution algorithm based on generative adversarial networks

Liu Lin, Jiang Yu

Proceedings Volume 12599, Second International Conference on Digital Society and Intelligent Systems (DSInS 2022); 125991Z (2023) https://doi.org/10.1117/12.2673371
Event: 2nd International Conference on Digital Society and Intelligent Systems (DSInS 2022), 2022, Chendgu, China

Abstract

Aiming at the issues that the current Generative Adversarial Network (GAN) used for image super-resolution reconstruction is easy to produce false details and the mapping dependence is not flexible enough, a One-to-many Generative Adversarial Network (OTMGAN) based on Enhanced Super Resolution Generative Adversarial Network (ESRGAN) was proposed. First, the model uses one-to-many constraints to allow the estimated patches to dynamically seek the best supervision during training, instead of the one-to-one constraint from low-resolution (LR) images to high-resolution (HR) images, thus improving the loss function of the model. Then, a new region-aware adversarial learning strategy is used to distinguish between flat regions and textured regions, so that the model can adaptively generate more realistic details for textured regions during training. Finally, the experimental consequences of image super resolution reestablishment on the public datasets BSDS100, Set5, Set14 and DIV2K show that the new model is significantly superior to the classical models such as ESRGAN in terms of Peak Signal-to-Noise Ratio (PSNR), Structural Similarity (SSIM), and Learning Perception Image Patch Similarity (LPIPS).

Citation Download Citation

Liu Lin and Jiang Yu "One-to-many Image super resolution algorithm based on generative adversarial networks", Proc. SPIE 12599, Second International Conference on Digital Society and Intelligent Systems (DSInS 2022), 125991Z (3 April 2023); https://doi.org/10.1117/12.2673371

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available

Members: $17.00

Non-members: $21.00 ADD TO CART

PROCEEDINGS
8 PAGES

DOWNLOAD PAPER SAVE TO MY LIBRARY

GET CITATION

RIGHTS & PERMISSIONS

Get copyright permission Get copyright permission on Copyright Marketplace

KEYWORDS

Super resolution

Image quality

Image restoration

Adversarial training

Lawrencium

Radium

Image enhancement

Show All Keywords

Keywords/Phrases

Search In:

Publication Years