Light-field image super-resolution based on multi-scale feature fusion

Q3 Engineering 光电工程 Pub Date : 2020-12-22 DOI:10.12086/OEE.2020.200007
Z. Yuanyuan, Shi Shengxian
{"title":"Light-field image super-resolution based on multi-scale feature fusion","authors":"Z. Yuanyuan, Shi Shengxian","doi":"10.12086/OEE.2020.200007","DOIUrl":null,"url":null,"abstract":"As a new generation of the imaging device, light-field camera can simultaneously capture the spatial position and incident angle of light rays. However, the recorded light-field has a trade-off between spatial resolution and angular resolution. Especially the application range of light-field cameras is restricted by the limited spatial resolution of sub-aperture images. Therefore, a light-field super-resolution neural network that fuses multi-scale features to obtain super-resolved light-field is proposed in this paper. The deep-learning-based network framework contains three major modules: multi-scale feature extraction, global feature fusion, and up-sampling. Firstly, inherent structural features in the 4D light-field are learned through the multi-scale feature extraction module, and then the fusion module is exploited for feature fusion and enhancement. Finally, the up-sampling module is used to achieve light-field super-resolution. The experimental results on the synthetic light-field dataset and real-world light-field dataset showed that this method outperforms other state-of-the-art methods in both visual and numerical evaluations. In addition, the super-resolved light-field images were applied to depth estimation in this paper, the results illustrated that the disparity map was enhanced through the light-field spatial super-resolution.","PeriodicalId":39552,"journal":{"name":"光电工程","volume":"12 2.1 1","pages":"200007"},"PeriodicalIF":0.0000,"publicationDate":"2020-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"光电工程","FirstCategoryId":"1087","ListUrlMain":"https://doi.org/10.12086/OEE.2020.200007","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"Engineering","Score":null,"Total":0}
引用次数: 0

Abstract

As a new generation of the imaging device, light-field camera can simultaneously capture the spatial position and incident angle of light rays. However, the recorded light-field has a trade-off between spatial resolution and angular resolution. Especially the application range of light-field cameras is restricted by the limited spatial resolution of sub-aperture images. Therefore, a light-field super-resolution neural network that fuses multi-scale features to obtain super-resolved light-field is proposed in this paper. The deep-learning-based network framework contains three major modules: multi-scale feature extraction, global feature fusion, and up-sampling. Firstly, inherent structural features in the 4D light-field are learned through the multi-scale feature extraction module, and then the fusion module is exploited for feature fusion and enhancement. Finally, the up-sampling module is used to achieve light-field super-resolution. The experimental results on the synthetic light-field dataset and real-world light-field dataset showed that this method outperforms other state-of-the-art methods in both visual and numerical evaluations. In addition, the super-resolved light-field images were applied to depth estimation in this paper, the results illustrated that the disparity map was enhanced through the light-field spatial super-resolution.
查看原文
分享 分享
微信好友 朋友圈 QQ好友 复制链接
本刊更多论文
基于多尺度特征融合的光场图像超分辨率
光场相机作为新一代成像设备,可以同时捕捉光线的空间位置和入射角。然而,记录的光场需要在空间分辨率和角分辨率之间进行权衡。特别是光场相机的应用范围受到子孔径图像空间分辨率的限制。为此,本文提出了一种融合多尺度特征的光场超分辨神经网络来获得超分辨光场。基于深度学习的网络框架包含三个主要模块:多尺度特征提取、全局特征融合和上采样。首先通过多尺度特征提取模块学习四维光场的固有结构特征,然后利用融合模块进行特征融合和增强。最后,利用上采样模块实现光场超分辨率。在合成光场数据集和真实光场数据集上的实验结果表明,该方法在视觉和数值评估方面都优于其他最先进的方法。此外,本文还将超分辨光场图像应用于深度估计,结果表明,光场空间超分辨增强了视差图。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 去求助
来源期刊
光电工程
光电工程 Engineering-Electrical and Electronic Engineering
CiteScore
2.00
自引率
0.00%
发文量
6622
期刊介绍: Founded in 1974, Opto-Electronic Engineering is an academic journal under the supervision of the Chinese Academy of Sciences and co-sponsored by the Institute of Optoelectronic Technology of the Chinese Academy of Sciences (IOTC) and the Optical Society of China (OSC). It is a core journal in Chinese and a core journal in Chinese science and technology, and it is included in domestic and international databases, such as Scopus, CA, CSCD, CNKI, and Wanfang. Opto-Electronic Engineering is a peer-reviewed journal with subject areas including not only the basic disciplines of optics and electricity, but also engineering research and engineering applications. Optoelectronic Engineering mainly publishes scientific research progress, original results and reviews in the field of optoelectronics, and publishes related topics for hot issues and frontier subjects. The main directions of the journal include: - Optical design and optical engineering - Photovoltaic technology and applications - Lasers, optical fibres and communications - Optical materials and photonic devices - Optical Signal Processing
期刊最新文献
The joint discriminative and generative learning for person re-identification of deep dual attention Fiber coupling technology of high brightness blue laser diode A few-shot learning based generative method for atmospheric polarization modelling Characteristics of wavefront correction using stacked liquid lens based on electrowetting-on-dielectric Research on joint coding for underwater single-photon video communication
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
现在去查看 取消
×
提示
确定
0
微信
客服QQ
Book学术公众号 扫码关注我们
反馈
×
意见反馈
请填写您的意见或建议
请填写您的手机或邮箱
已复制链接
已复制链接
快去分享给好友吧!
我知道了
×
扫码分享
扫码分享
Book学术官方微信
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术
文献互助 智能选刊 最新文献 互助须知 联系我们:info@booksci.cn
Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。
Copyright © 2023 Book学术 All rights reserved.
ghs 京公网安备 11010802042870号 京ICP备2023020795号-1