计算机科学 ›› 2021, Vol. 48 ›› Issue (8): 139-144.doi: 10.11896/jsjkx.200500150

• 计算机图形学& 多媒体 • 上一篇    下一篇

基于通道注意递归残差网络的图像超分辨率重建

郭琳1,2,3, 李晨1, 陈晨1, 赵睿1, 范仕霖1, 徐星雨1   

  1. 1 湖北大学计算机与信息工程学院 武汉430062
    2 智慧政务与人工智能应用湖北省工程研究中心 武汉430062
    3 湖北省教育信息化工程技术研究中心 武汉430062
  • 收稿日期:2020-05-28 修回日期:2020-09-17 发布日期:2021-08-10
  • 通讯作者: 郭琳(guolin@hubu.edu.cn)
  • 基金资助:
    国家自然科学基金(61806076);湖北省自然科学基金项目(2018CFB158)

Image Super-resolution Reconstruction Using Recursive ResidualNetwork Based on ChannelAttention

GUO Lin1,2,3, LI Chen1, CHEN Chen1, ZHAO Rui1, FAN Shi-lin1, XU Xing-yu1   

  1. 1 School of Computer Science and Information Engineering,Hubei University,Wuhan 430062,China;
    2 Hubei Provincial Engineering Research Center for Smart Government Affairs and Artificial Intelligence Application,Wuhan 430062,China;
    3 Research Center of Educational Informatization Engineering and Technology,Hubei Province,Wuhan 430062,China
  • Received:2020-05-28 Revised:2020-09-17 Published:2021-08-10
  • About author:GUO Lin,born in 1978,Ph.D,associate professor.Her main research interests include signal processing,machine vision and deep learning.
  • Supported by:
    National Natural Science Foundation of China(61806076) and Hubei Provincial Natural Science Foundation of China(2018CFB158).

摘要: 近年来,深度学习被广泛应用于图像超分辨率重建。针对基于深度学习的超分辨率重建方法存在的特征提取不充分、细节丢失和梯度消失等问题,提出一种基于通道注意的递归残差深度神经网络模型,用于单幅图像的超分辨率重建。该模型采用残差嵌套网络和跳跃连接构成一种简洁的递归残差网络结构,能够加快深层网络的收敛,同时避免网络退化和梯度问题。在特征提取部分,引入注意力机制来提升网络的判别性学习能力,以提取到更准确、有效的深层残差特征;随后结合并行映射重建网络,最终实现超分辨率重建。在数据集Set5,Set14,B100和Urban100上进行放大2倍、3倍和4倍的重建测试实验,并从客观指标和主观视觉效果上将所提方法与主流方法进行比较。实验结果显示,所提方法在全部4个测试数据集上的客观指标较对比方法均有明显提升,其中,相比插值法和SRCNN 算法,在放大2倍、3倍、4倍时所提方法的平均PSNR值分别提升了3.965dB和1.56dB、3.19dB和1.42dB、2.79dB和1.32dB。视觉效果对比也表明所提方法能更好地恢复图像细节。

关键词: 超分辨率, 深度学习, 通道注意, 残差网络, 跳跃连接

Abstract: In recent years,deep learning has been widely used in image super-resolution reconstruction.To solve the problems of inadequate feature extraction,loss of details and gradient disappearance in super-resolution reconstruction methods based on deep learning,a deep recursive residual neural network model based on channel attention is proposed for single image super-resolution reconstruction.The proposed model constructs a simple recursive residual network structure by residual nested networks and jump connections to deepen the network and speed up its convergence while avoiding network degradation and gradient problems.An attention mechanism is introduced into the feature extraction part to improve the discriminant learning ability of the network for more accurate and more effective extraction of deep residual features,which is combined with the subsequent reconstruction network with parallel mapping structure to ensure the final accurate reconstruction.Quantitative and qualitative assessments are performed on benchmark dataset Set5,Set14,B100 and Urban100 at the magnification of 2,3 and 4 times by comparison with the mainstream methods.Experimental results show that the objective index values of the proposed method increase significantly compared to the comparative methods on all four test data sets.Among them,compared with the interpolation method and the SRCNN algorithm,the average PSNR improves 3.965dB and 1.56dB,3.19dB and 1.42dB,2.79dB and 1.32dB,respectively,at the magnification of 2,3 and 4 times.Visual effects show that the proposed method can recover image details better.

Key words: Super-resolution, Deep learning, Channel attention, Residual network, Skip connections

中图分类号: 

  • TP391
[1]ZHANG H,ZHANG L,SHEN H.A blind super-resolution reconstruction method considering image registration errors [J].International Journal of Fuzzy Systems,2015,17(2):353-364.
[2]ZHANG K B,GAO X B,TAO D A,et al.Single image super-resolution with non-local means and steering kernel regression[J].IEEE Transactions on Image Processing,2012,21( 11):4544-4556.
[3]TIMOFTE R,DE V,GOOL L V.Anchored neighborhood regression for fast example-based super-resolution[C]//Procee-dings of 2013 IEEE International Conference on Computer Vision.Sydney,NSW,Australia:IEEE,2013:1920-1927.
[4]TIMOFTE R,DESMET V,VANGOOL L.A+:adjusted an-chored neighborhood regression for fast super-resolution[C]//Proceedings of 12th Asian Conference on Computer Vision.Singapore:Springer,2014:111-126.
[5]PELEG T,ELAD M.A statistical prediction model based onsparse representations for single image super-resolution [J].IEEE Transactions on Image Processing,2014,23(6):2569-2582.
[6]QIN X J,SHAN Y Y,XIAO J J,et al.Self-learning single image super-resolution reconstruction based on compressive sensing and SVR[J].Computer Science,2017,44(S2):169-174.
[7]LI J H,WU Y R,LV J J.Onlinesingle image super-resolution algorithm based on group sparse representation[J].Computer Science,2018,45(4):312-318.
[8]THAPA D,RAAHEMIFAR K,BOBIER W R,et al.A perfor-mance comparison among different super-resolution techniques [J].Computers & Electrical Engineering,2016,54:313-329.
[9]DONG C,LOY C C,HE K M,et al.Image super-resolutionusing deep convolutional networks[J].IEEE Transaction on Pattern Analysis & Machine Intelligence,2016,38(2):295-307.
[10]DONG C,CHEN C L,TANG X.Accelerating the Super-Reso-lution Convolutional Neural Network[C]//European Conference on Computer Vision.Springer,Cham,2016:391-407.
[11]KIM J,JUNG K L,KYOUNG M L.Accurate Image Super-Re-solution Using Very Deep Convolutional Networks[J].IEEE Conference on Computer Vision and Pattern Recognition.2016:1646-1654.
[12]KIM J,JUNG K L,KYOUNG M L.Deeply-recursive convolutional network for image super-resolution[C]//IEEE Confe-rence on Computer Vision and Pattern Recognition.2016:1637-1645.
[13]LAI W S,HUANG J B,NARENDRA A,et al.Deep laplacian pyramid networks for fast and accurate super-resolution[C]//IEEE Conference on Computer Vision and Pattern Recognition.2017:624-632.
[14]QU Y,LIN L,SHEN F,et al.Joint Hierarchical Category Structure Learning and Large-Scale Image Classification[J]IEEE Transactions on Image Processing,2017,26(9):4331-4346.
[15]HE K,ZHANG X,REN S,et al.Deep Residual Learning for Ima-ge Recognition[C]//IEEE Conference on Computer Vision and Pattern Recognition.2016:770-778.
[16]SHI W Z,CABALLERO J,HUSZAR F,et al.Real-time single image and video super-resolution using an effcient sub-pixel convolutional neural network[C]//IEEE Conference on Computer Vision and Pattern Recognition.2016:1874-1883.
[17]TAI Y,YANG J,LIU X M.Image super-resolution via deep recursive residual network [C]//IEEE Conference on Computer Vision and Pattern Recognition.2017:3147-3155.
[18]TAI Y,YANG J,LIU X M.Memnet:A persistent memory network for image restoration[C]//IEEE International Conference on Computer Vision.2017:4539-4547.
[19]HU J,LI S,ALBANIE S,et al.Squeeze-and-Excitation Net-works[C]//IEEE Conference on Computer Vision and Pattern Recognition.2018:7132-7141.
[20]HE K,ZHANG X,REN S,et al.Delving deep into rectifiers:Surpassing human-level performance on imagenet classification [C]//Proceedings of the IEEE International Conference on Computer Vision Santiago.Chile,2015:1026-1034.
[21]QIN Z S,ZHU L L,ZANG H J.An image super-resolution reconstruction method based on convolutional neural network:China Patent,201910149271.6[P].2019-06-18[2020-09-17].
[1] 冯霞, 胡志毅, 刘才华. 跨模态检索研究进展综述[J]. 计算机科学, 2021, 48(8): 13-23.
[2] 周文辉, 石敏, 朱登明, 周军. 基于残差注意力网络的地震数据超分辨率方法[J]. 计算机科学, 2021, 48(8): 24-31.
[3] 王立梅, 朱旭光, 汪德嘉, 张勇, 邢春晓. 基于深度学习的民事案件判决结果分类方法研究[J]. 计算机科学, 2021, 48(8): 80-85.
[4] 刘帅, 芮挺, 胡育成, 杨成松, 王东. 基于深度学习SuperGlue算法的单目视觉里程计[J]. 计算机科学, 2021, 48(8): 157-161.
[5] 王施云, 杨帆. 基于U-Net特征融合优化策略的遥感影像语义分割方法[J]. 计算机科学, 2021, 48(8): 162-168.
[6] 许华杰, 张晨强, 苏国韶. 基于深层卷积残差网络的航拍图建筑物精确分割方法[J]. 计算机科学, 2021, 48(8): 169-174.
[7] 田嵩旺, 蔺素珍, 杨博. 基于多判别器的多波段图像自监督融合方法[J]. 计算机科学, 2021, 48(8): 185-190.
[8] 潘孝勤, 芦天亮, 杜彦辉, 仝鑫. 基于深度学习的语音合成与转换技术综述[J]. 计算机科学, 2021, 48(8): 200-208.
[9] 汤世征, 张岩峰. DragDL:一种易用的深度学习模型可视化构建系统[J]. 计算机科学, 2021, 48(8): 220-225.
[10] 张瑾, 段利国, 李爱萍, 郝晓燕. 基于注意力与门控机制相结合的细粒度情感分析[J]. 计算机科学, 2021, 48(8): 226-233.
[11] 刘文洋, 郭延哺, 李维华. 识别关键蛋白质的混合深度学习模型[J]. 计算机科学, 2021, 48(8): 240-245.
[12] 王超, 魏祥麟, 田青, 焦翔, 魏楠, 段强. 基于特征梯度的调制识别深度网络对抗攻击方法[J]. 计算机科学, 2021, 48(7): 25-32.
[13] 羊洋, 陈伟, 张丹懿, 王丹妮, 宋爽. 对抗攻击威胁基于卷积神经网络的网络流量分类[J]. 计算机科学, 2021, 48(7): 55-61.
[14] 暴雨轩, 芦天亮, 杜彦辉, 石达. 基于i_ResNet34模型和数据增强的深度伪造视频检测方法[J]. 计算机科学, 2021, 48(7): 77-85.
[15] 桑春艳, 胥文, 贾朝龙, 文俊浩. 社交网络中基于注意力机制的网络舆情事件演化趋势预测[J]. 计算机科学, 2021, 48(7): 118-123.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
[1] 孙启,金燕,何琨,徐凌轩. 用于求解混合车辆路径问题的混合进化算法[J]. 计算机科学, 2018, 45(4): 76 -82 .
[2] 施超,谢在鹏,柳晗,吕鑫. 基于稳定匹配的容器部署策略的优化[J]. 计算机科学, 2018, 45(4): 131 -136 .
[3] 崔璐,张鹏,车进. 基于深度神经网络的遥感图像分类算法综述[J]. 计算机科学, 2018, 45(6A): 50 -53 .
[4] 余伟伟,谢承旺. 一种多策略混合的粒子群优化算法[J]. 计算机科学, 2018, 45(6A): 120 -123 .
[5] 付泰,杨力,王斌. 一种具有精确位姿的飞机CAD模型投影图像库设计方法[J]. 计算机科学, 2018, 45(6A): 244 -246 .
[6] 张滨, 乐嘉锦. 基于列存储的MapReduce分布式Hash连接算法[J]. 计算机科学, 2018, 45(6A): 471 -475 .
[7] 曾新,李晓伟,杨健. 基于数据规范化的co-location模式挖掘算法[J]. 计算机科学, 2018, 45(6A): 482 -486 .
[8] 李运筹,尹平. 模型检验在航天测控软件上的应用研究[J]. 计算机科学, 2018, 45(6A): 523 -526 .
[9] 薛艳, 武淑红, 王耀力. 基于SPIN的G语言系统模型的验证[J]. 计算机科学, 2018, 45(6A): 536 -540 .
[10] 吴鹏, 周杰, 陈姜高路. SOC水声信道模型及其计算方法研究[J]. 计算机科学, 2018, 45(8): 94 -99 .