G-GANISR: Gradual generative adversarial network for image super resolution

doi:10.1016/j.neucom.2019.07.094

	G-GANISR: Gradual generative adversarial network for image super resolution
	Shamsolmoali, Pourya 1; Zareapoor, Masoumeh 1,2; Wang, Ruili 3; Jain, Deepak Kumar 4; Yang, Jie 1
刊名	NEUROCOMPUTING
	2019-11-13
卷号	366 页码:140-153
关键词	Image super-resolution Loss functions GAN CNN Gradual learning
ISSN号	0925-2312
DOI	10.1016/j.neucom.2019.07.094
通讯作者	Wang, Ruili(ruili.wang@massey.ac.nz)
英文摘要	Adversarial methods have demonsterated to be signifiant at generating realistic images. However, these approaches have a challenging training process which partially attributed to the performance of discriminator. In this paper, we proposed an efficient super-resolution model based on generative adversarial network (GAN), to effectively generate reprehensive information and improve the test quality of the real-world images. To overcome the current issues, we designed the discriminator of our model based on the Least Square Loss function. The proposed network is organized by a gradual learning process from simple to advanced, which means from the small upsampling factors to the large upsampling factor that helps to improve the overall stability of the training. In particular, to control the model parameters and mitigate the training difficulties, dense residual learning strategy is adopted. Indeed, the key idea of proposed methodology is (i) fully exploit all the image details without losing information by gradually increases the task of discriminator, where the output of each layer is gradually improved in the next layer. In this way the model efficiently generates a super-resolution image even up to high scaling factors (e.g. x 8). (ii) The model is stable during the learning process, as we use least square loss instead of cross-entropy. In addition, the effects of different objective function on training stability are compared. To evaluate the model we conducted two sets of experiments, by using the proposed gradual GAN and the regular GAN to demonstrate the efficiency and stability of the proposed model for both quantitative and qualitative benchmarks. (C) 2019 Published by Elsevier B.
WOS关键词	SUPERRESOLUTION
WOS研究方向	Computer Science
语种	英语
出版者	ELSEVIER
WOS记录号	WOS:000488202500014
内容类型	期刊论文
源URL	[http://ir.ia.ac.cn/handle/173211/26629]
专题	中国科学院自动化研究所
通讯作者	Wang, Ruili
作者单位	1.Shanghai Jiao Tong Univ, Inst Image Proc & Pattern Recognit, Shanghai, Peoples R China 2.Tokyo Univ Technol, Dept Comp Sci, Tokyo, Japan 3.Massey Univ, Inst Nat & Math Sci, Auckland, New Zealand 4.Chinese Acad Sci, Inst Automat, Beijing, Peoples R China
推荐引用方式 GB/T 7714	Shamsolmoali, Pourya,Zareapoor, Masoumeh,Wang, Ruili,et al. G-GANISR: Gradual generative adversarial network for image super resolution[J]. NEUROCOMPUTING,2019,366:140-153.
APA	Shamsolmoali, Pourya,Zareapoor, Masoumeh,Wang, Ruili,Jain, Deepak Kumar,&Yang, Jie.(2019).G-GANISR: Gradual generative adversarial network for image super resolution.NEUROCOMPUTING,366,140-153.
MLA	Shamsolmoali, Pourya,et al."G-GANISR: Gradual generative adversarial network for image super resolution".NEUROCOMPUTING 366(2019):140-153.