基于可嵌入式网络结构的图像超分辨率重建方法

doi:10.19678/j.issn.1000-3428.0056980

计算机工程 ›› 2021, Vol. 47 ›› Issue (5): 221-228. doi: 10.19678/j.issn.1000-3428.0056980

基于可嵌入式网络结构的图像超分辨率重建方法

强保华^1,2, 庞远超^1,2, 杨明浩¹, 曾坤⁴, 郑虹¹, 谢武^2,3, 莫烨^1,2

1. 桂林电子科技大学广西图像图形与智能处理重点实验室, 广西桂林 541004;
2. 桂林电子科技大学广西可信软件重点实验室, 广西桂林 541004;
3. 桂林电子科技大学广西云计算与大数据协同创新中心, 广西桂林 541004;
4. 闽江大学福建省信息处理与智能控制重点实验室, 福州 350108

收稿日期:2019-12-20 修回日期:2020-02-14 发布日期:2020-04-22
作者简介:强保华(1972-),男,教授、博士,主研方向为图像处理、大数据分析;庞远超,硕士研究生;杨明浩,副研究员、博士;曾坤、郑虹(通信作者),讲师、博士;谢武,副教授、博士;莫烨,硕士。
基金资助:
国家自然科学基金（61762025）；国家自然科学基金青年基金（61601389）；广西重点研发计划（桂科AB17195053，桂科AB18126053）；广西自然科学基金（2017GXNSFAA198226）；桂林科技发展计划（20180107-4）。

Image Super-Resolution Reconstruction Method Based on Embeddable Network Structure

QIANG Baohua^1,2, PANG Yuanchao^1,2, YANG Minghao¹, ZENG Kun⁴, ZHENG Hong¹, XIE Wu^2,3, MO Ye^1,2

1. Guangxi Key Laboratory of Image Graphics and Intelligent Processing, Guilin University of Electronic Technology, Guilin, Guangxi 541004, China;
2. Guangxi Key Laboratory of Trusted Software, Guilin University of Electronic Technology, Guilin, Guangxi 541004, China;
3. Guangxi Cloud Computing and Big Data Collaborative Innovation Center, Guilin University of Electronic Technology, Guilin, Guangxi 541004, China;
4. Fujian Provincial Key Laboratory of Information Processing and Intelligent Control, Minjiang University, Fuzhou 350108, China

Received:2019-12-20 Revised:2020-02-14 Published:2020-04-22

摘要/Abstract

摘要： 针对卷积神经网络中的图像超分辨率重建模型训练不稳定与收敛速度较慢等问题，提出一种可嵌入式并行网络框架（EPNF），用于单幅图像超分辨率重建任务。将现有的图像超分辨率网络模型作为EPNF框架的深层结构部分嵌入到该框架中，能够以较小参数代价加快所嵌入的超分辨率模型的收敛速度，在一定程度上提高模型的准确率。在EPNF网络结构的基础上，提出一种新的超分辨率重建方法EPNF_DCSR，采用稠密跳跃连接构造高分辨率（HR）图像的高频成分，并使用单层卷积构造HR图像的低频成分，合成一幅HR输出图像。实验结果表明，与当前主流的图像超分辨率算法相比，EPNF_DCSR具有更好的图像生成效果。

关键词: 卷积神经网络, 上采样, 并行网络, 跳跃连接, 图像超分辨率

Abstract: The existing Super-Resolution(SR) image reconstruction models based on Convolutional Neural Networks(CNN) have multiple deficiencies,such as instable model training process and low convergence speed.To address the problem,this paper proposes an Embeddable Parallel Network Framework(EPNF) for the super-resolution reconstruction of a single image.The existing network models for super-resolution image reconstruction can be embedded into EPNF as the deep structure part of the framework,improving their convergence speed and accuracy by taking a small loss of parameters.On the basis of EPNF,this paper also proposes a novel super-resolution image reconstruction method,EPNF_DCSR.The method uses the dense jump connections to restore the high-frequency details of high-resolution images,and a single-layer convolution to restore low-frequency details.Finally a high-resolution image is output.Experimental results show that EPNF_DCSR has better image reconstruction performance than the existing mainstream image reconstruction algorithms.

Key words: Convolutional Neural Networks(CNN), upsampling, parallel network, skip connection, image Super-Resolution(SR)

中图分类号:

TP391

强保华, 庞远超, 杨明浩, 曾坤, 郑虹, 谢武, 莫烨. 基于可嵌入式网络结构的图像超分辨率重建方法[J]. 计算机工程, 2021, 47(5): 221-228.

QIANG Baohua, PANG Yuanchao, YANG Minghao, ZENG Kun, ZHENG Hong, XIE Wu, MO Ye. Image Super-Resolution Reconstruction Method Based on Embeddable Network Structure[J]. Computer Engineering, 2021, 47(5): 221-228.

https://www.ecice06.com/CN/Y2021/V47/I5/221

参考文献

[1] DONG C,LOY C C,HE K,et al.Image super-resolution using deep convolutional networks[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2016,38(2):295-307.
[2] DONG C,LOY C C,TANG X.Accelerating the super-resolution convolutional neural network[C]//Proceedings of European Conference on Computer Vision.Berlin,Germany:Springer,2016:391-407.
[3] SHI W Z,CABALLERO J,HUSZÁR F,et al.Real-time single image and video super-resolution using an efficient sub-pixel convolutional neural network[EB/OL].[2019-11-10].https://arxiv.org/abs/1609.05158.
[4] KIM J,LEE J K,LEE K M.Accurate image super-resolution using very deep convolutional networks[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C.,USA:IEEE Press,2016:1646-1654.
[5] LEDIG C,THEIS L,HUSZAR F,et al.Photo-realistic single image super-resolution using a generative adversarial network[EB/OL].[2019-11-10].https://arxiv.org/abs/1609.04802.
[6] LIM B,SON S,KIM H,et al.Enhanced deep residual networks for single image super-resolution[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C.,USA:IEEE Press,2017:312-124.
[7] TONG Tong,LI Gen,LIU Xiejie,et al.Image super-resolution using dense skip connections[C]//Proceedings of IEEE International Conference on Computer Vision.[S.1.]:IEEE Computer Society,2017:4809-4817.
[8] HUANG G,LIU Z,WEINBERGER K Q,et al.Densely connected convolutional networks[EB/OL].[2019-11-10].https://arxiv.org/abs/1608.06993.
[9] WANG Yifan,WANG Liujun,WANG Hongyu,et al.End-to-end image super-resolution via deep and shallow convolutional networks[EB/OL].[2019-11-10].https://arxiv.org/abs/1607.07680.
[10] YU Jiahui,FAN Yichneg,YANG Jianchao,et al.Wide activation for efficient and accurate image super-resolution[EB/OL].[2019-11-10].https://arxiv.org/abs/1808.08718.
[11] LECUN Y,BOTTOU L,BENGIO Y,et al.Gradient-based learning applied to document recognition[J].Proceedings of the IEEE,1998,86(11):2278-2324.
[12] KRIZHEVSKY A,SUTSKEVER I,HINTON G.ImageNet classification with deep convolutional neural networks[C]//Proceedings of NIPS'12.Cambrigde,USA:MIT Press,2012:369-378.
[13] SIMONYAN K,ZISSERMAN A.Very deep convolutional networks for large-scale image recognition[EB/OL].[2019-11-10].https://arxiv.org/abs/.
[14] SZEGEDY C,LIU W,JIA Y,et al.Going deeper with convolutions[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C.,USA:IEEE Press,2015:458-467.
[15] HE Kaiming,ZHANG Xiangyu,REN Shaoqing,et al.Deep residual learning for image recognition[EB/OL].[2019-11-10].https://arxiv.org/abs/1512.03385.
[16] TAI Ying,YANG Jian,LIU Xiaoming.Image super-resolution via deep recursive residual network[C]//Proceedings of IEEE Computer Vision and Pattern Recognition.Washington D.C.,USA:IEEE Press,2017:356-368.
[17] DODGSON N A.Quadratic interpolation for image resampling[J].IEEE Transactions on Image Processing,1997,6(9):1322-1326.
[18] ODENA A,DUMOULIN V,OLAH C.Deconvolution and checkerboard artifacts[EB/OL].[2019-11-10].https://distill.pub/2016/deconv-checkerboard/.
[19] LUGMAYR A,DANELLJAN M,TIMOFTE R,et al.NTIRE 2020 challenge on real-world image super-resolution:methods and results[EB/OL].[2019-11-10].https://arxiv.org/abs/2005.01996.
[20] BEVILACQUA M,ROUMY A,GUILLEMOT C,et al.Low-complexity single-image super-resolution based on nonnegative neighbor embedding[C]//Proceedings of IEEE BMVC'12.Washington D.C.,USA:IEEE Press,2012:456-468.
[21] ZEYDE R,ELAD M,PROTTER M.On single image scale-up using sparse-representations[C]//Proceedings of IEEE International Conference on Curves and Surfaces.Washington D.C.,USA:IEEE Press,2010:243-258.
[22] MARTIN D,FOWLKES C,TAL D,et al.A database of human segmented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics[C]//Proceedings of IEEE ICCV'01.Washington D.C.,USA:IEEE Press,2001:1125-1136.
[23] HUNAG J B,SINGH A,AHUJA N.Single image super-resolution from transformed self-exemplars[C]//Proceedings of CVPR'15.Washington D.C.,USA:IEEE Press,2015:335-348.
[24] TIMOFTE R,DE SMET V,VAN GOOL L.A+:adjusted anchored neighborhood regression for fast super-resolution[C]//Proceedings of International Conference on Computer Vision.Berlin,Germany:Springer,2014:111-126.

选择文件类型/文献管理软件名称

选择包含的内容

基于可嵌入式网络结构的图像超分辨率重建方法

Image Super-Resolution Reconstruction Method Based on Embeddable Network Structure

RichHTML

PDF

可视化

被引次数

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

本文评价

[1]	耿丽丽, 牛保宁. 基于通道相似度熵的卷积神经网络裁剪[J]. 计算机工程, 2024, 50(7): 133-143.
[2]	张洋, 刘畅, 李少青. 基于可控制性度量的图神经网络门级硬件木马检测方法[J]. 计算机工程, 2024, 50(7): 164-173.
[3]	牛瑞婷, 严天峰, 高锐, 王映植. 低信噪比下基于深度学习TCNN-MobileNet的调制识别[J]. 计算机工程, 2024, 50(7): 204-215.
[4]	张溢文, 蔡满春, 陈咏豪, 朱懿, 姚利峰. 融合空间特征的多尺度深度伪造检测方法[J]. 计算机工程, 2024, 50(7): 240-250.
[5]	逯焕宇, 张永宏, 马光义, 谢东林, 田伟. 基于半监督对抗学习的遥感图像水体提取[J]. 计算机工程, 2024, 50(7): 251-263.
[6]	于洋, 孙芳芳, 吕华, 李扬, 王晓民. 基于多尺度时空注意力网络的微表情检测方法[J]. 计算机工程, 2024, 50(6): 228-235.
[7]	张雷, 沈国琛, 欧冬秀. 用于热成像数据的卷积神经网络特征图筛选方法[J]. 计算机工程, 2024, 50(4): 31-40.
[8]	张雷, 沈国琛, 欧冬秀. 用于热成像数据的卷积神经网络特征图筛选方法[J]. 计算机工程, 2024, 50(4): 31-40.
[9]	李政学, 李枝名, 彭德中, 陈杰. 基于特征对比学习和图卷积的社交网络用户分类[J]. 计算机工程, 2024, 50(4): 258-266.
[10]	姜百浩, 刘静, 仇大伟, 姜良. 深度学习在脊柱图像分割中的应用综述[J]. 计算机工程, 2024, 50(3): 1-15.
[11]	王柏涵, 姜晓燕, 范柳伊. 基于深度监督隐空间构建的语义分割改进方法[J]. 计算机工程, 2024, 50(3): 191-199.
[12]	谢新林, 尹东旭, 张涛源, 谢刚. 基于注意力机制的多尺度融合人群计数算法[J]. 计算机工程, 2024, 50(3): 290-297.
[13]	赵佳圆, 张玉茹, 苏晓东, 徐红岩, 李世洲, 张玉荣. 基于注意力机制的人体关键点隐式建模网络[J]. 计算机工程, 2024, 50(3): 317-325.
[14]	徐浩宸, 刘满华. 基于多层次自注意力网络的人脸特征点检测[J]. 计算机工程, 2024, 50(2): 239-246.
[15]	李倩, 魏伟波, 杨光宇, 宋金涛, 孙璐, 潘振宽. 基于先验驱动深度神经网络的泊松去噪变分模型[J]. 计算机工程, 2024, 50(2): 273-280.

模态框（Modal）标题

选择文件类型/文献管理软件名称

选择包含的内容

基于可嵌入式网络结构的图像超分辨率重建方法

Image Super-Resolution Reconstruction Method Based on Embeddable Network Structure

RichHTML

PDF

可视化

被引次数

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

本文评价