基于四路多尺度注意力模块及桥连结构的轻量级单图超分辨率网络

doi:10.16451/j.cnki.issn1003-6059.202511004

摘要
图/表
参考文献(0)
相关文章 (15)

全文: PDF (1482 KB) HTML (1 KB)
输出: BibTeX | EndNote (RIS)

摘要

针对以压缩-激励(Squeeze-and-Excitation, SE)注意力模块为主要组成的图像超分辨率网络A²F-SD的注意力机制过于简单和多路信息利用能力的不足,文中提出基于四路多尺度注意力模块及桥连结构的轻量级单图超分辨率网络(Lightweight Single-Image Super-Resolution Network Based on Four-Path Multi-scale Attention and Bridge Structure, A²F-MSAB),主要优化A²F中的SE注意力模块与沙漏型结构.首先,设计多尺度自学习融合均值与方差的注意力模块,分成四路增强网络对图像特征的提取能力.然后,将A²F-SD中的沙漏型结构改为桥连多级沙漏型结构,促进网络不同模块之间、浅层与深层之间的信息流动,增强对中高频细节的重建能力.实验表明,A²F-MSAB性能较优,在部分数据集上的指标值甚至优于A²F-SD及A²F-S.

	服务

	把本文推荐给朋友
	加入我的书架
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	苏博何俊
	徐勇
	薛睿
	刘伟
	王好谦

关键词 ：图像超分辨率, 注意力机制, 特征融合, 深度学习

Abstract：

A²F-SD, the image super-resolution network, is built on squeeze and excitation attention module. In this paper, the limitations of A²F-SD are analyzed, including the excessively simplistic attention mechanism and insufficient utilization of multi-path information. To address these issues, a lightweight single-image super-resolution network based on four-path multi-scale attention module and bridging structure(A²F-MSAB) is proposed to optimize the attention modules and hourglass structure in the original A²F-SD model. First, multi-scale self-learning average-variance attention modules are designed. A four-path attention module is employed to enhance the image feature extraction ability of the network. Then, the hourglass structure of A²F-SD is improved into the bridge and multi-level hourglass structure to facilitate information flow between different modules, and between shallow and deep layers of the model, thereby strengthening the model reconstruction ability for medium and high-frequency details. Experiments show that A²F-MSAB model with only half of the basic modules stacked achieves superior performance and its evaluation metrics on certain datasets outperform those of both A²F-SD and A²F-S.

Key words： Image Super-Resolution Attention Mechanism Feature Fusion Deep Learning

收稿日期: 2025-09-23

ZTFLH:

TP391

基金资助:

深圳市科技计划项目(No.KJZD20230923114600002)资助

通讯作者: 徐勇,博士,教授,主要研究方向为模式识别、生物特征识别、图像处理、视频分析、机器学习.E-mail:laterfall@hit.edu.cn.

作者简介: 苏博何俊,硕士研究生,主要研究方向为计算机视觉.E-mail:3389370532@qq.com.
薛睿,硕士,高级实验师,主要研究方向为计算机系统、计算机视觉.E-mail:xuerui@hit.edu.cn.
刘伟,硕士,讲师,主要研究方向为计算机视觉.E-mail:liuwei2020@hit.edu.cn.
王好谦,博士,教授,主要研究方向为计算摄像、人工智能.E-mail:wangyizhai@sz.tsinghua.edu.cn.

引用本文:

苏博何俊, 徐勇, 薛睿, 刘伟, 王好谦. 基于四路多尺度注意力模块及桥连结构的轻量级单图超分辨率网络[J]. 模式识别与人工智能, 2025, 38(11): 999-1012. SU Bohejun, XU Yong, XUE Rui, LIU Wei, WANG Haoqian. Lightweight Single-Image Super-Resolution Network Based on Four-Path Multi-scale Attention Module and Bridging Structure. Pattern Recognition and Artificial Intelligence, 2025, 38(11): 999-1012.

链接本文:

http://manu46.magtech.com.cn/Jweb_prai/CN/10.16451/j.cnki.issn1003-6059.202511004 或 http://manu46.magtech.com.cn/Jweb_prai/CN/Y2025/V38/I11/999

[1] DONG C, LOY C C, HE K M, et al. Image Super-Resolution Using Deep Convolutional Networks. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 38(2): 295-307.
[2] ANWAR S, KHAN S, BARNES N.A Deep Journey into Super-Re-solution: A Survey. ACM Computing Surveys, 2020, 53(3). DOI: 10.1145/339046.
[3] HU J, SHEN L, ALBANIE S, et al. Squeeze-and-Excitation Networks. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2020, 42(8): 2011-2023.
[4] WANG X H, WANG Q, ZHAO Y Z, et al. Lightweight Single Image Super-Resolution Network with Attentive Auxiliary Feature Learning // Proc of the 15th Asian Conference on Computer Vision. Berlin, Germany: Springer, 2020: 268-285.
[5] WOO S, PARK J, LEE J, et al. CBAM: Convolutional Block Attention Module // Proc of the 15th European Conference on Computer Vision. Berlin, Germany: Springer, 2018: 3-19.
[6] HUI Z, GAO X B, WANG X M.Lightweight Image Super-Resolution with Feature Enhancement Residual Network. Neurocomputing, 2020, 404: 50-60.
[7] MA H L, CHU X X, ZHANG B.Accurate and Efficient Single Image Super-Resolution with Matrix Channel Attention Network // Proc of the 15th Asian Conference on Computer Vision. Berlin, Germany: Springer, 2020: 20-35.
[8] 施举鹏,李静,陈琰,等.DFAN:一种基于深度反馈注意力网络的图像超分辨率方法.小型微型计算机系统, 2021, 42(6): 1206-1212.
(SHI J P, LI J, CHEN Y, et al. DFAN: A Deep Feedback Attention Network for Image Super-Resolution. Journal of Chinese Computer Systems, 2021, 42(6): 1206-1212.)
[9] HANG Y C, LIAO Q M, YANG W M, et al. Attention Cube Network for Image Restoration // Proc of the 28th ACM International Conference on Multimedia. New York, USA: ACM, 2020: 2562-2570.
[10] 李方玗,贾晓芬,赵佰亭,等.高效多注意力特征融合的图像超分辨率重建算法.小型微型计算机系统, 2023, 44(5): 1023-1028.
(LI F Y, JIA X F, ZHAO B T, et al. Efficient Multi-attention Feature Fusion for Image Super-Resolution Reconstruction Algorithms. Journal of Chinese Computer Systems, 2023, 44(5): 1023-1028.)
[11] 李千,赵逢禹.融合多路径与混合注意力的遥感图像超分辨率重建.小型微型计算机系统, 2023, 44(7): 1508-1513.
(LI Q, ZHAO F Y.Super-Resolution of Remote Sensing Images Based on Multi-path and Mixed Attention. Journal of Chinese Computer Systems, 2023, 44(7): 1508-1513.)
[12] YAN Y T, XU X, CHEN W H, et al. Lightweight Attended Multi-scale Residual Network for Single Image Super-Resolution. IEEE Access, 2021, 9: 52202-52212.
[13] YU J H, FAN Y C, YANG J C, et al. Wide Activation for Efficient and Accurate Image Super-Resolution[C/OL].[2025-07-25]. https://arxiv.org/pdf/1808.08718.
[14] REDMON J, DIVVALA S, GIRSHICK R, et al. You Only Look Once: Unified, Real-Time Object Detection // Proc of the IEEE Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2016: 779-788.
[15] AGUSTSSON E, TIMOFTE R.NTIRE 2017 Challenge on Single Image Super-Resolution: Dataset and Study // Proc of the IEEE Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2017: 1122-1131.
[16] DONG C, LOY C C, TANG X O.Accelerating the Super-Resolution Convolutional Neural Network // Proc of the 14th European Conference on Computer Vision. Berlin, Germany: Springer, 2016: 391-407.
[17] WANG Z W, LIU D, YANG J C, et al. Deep Networks for Image Super-Resolution with Sparse Prior // Proc of the IEEE International Conference on Computer Vision. Washington, USA: IEEE, 2015: 370-378.
[18] LAI W S, HUANG J B, AHUJA N, et al. Deep Laplacian Pyramid Networks for Fast and Accurate Super-Resolution // Proc of the IEEE Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2017: 5835-5843.
[19] SHOCHER A, COHEN N, IRANI M.Zero-Shot Super-Resolution Using Deep Internal Learning // Proc of the IEEE/CVF Confe-rence on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2018: 3118-3126.
[20] KIM J, LEE J K, LEE K M.Deeply-Recursive Convolutional Network for Image Super-Resolution // Proc of the IEEE Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2016: 1637-1645.
[21] TAI Y, YANG J, LIU X M.Image Super-Resolution via Deep Recursive Residual Network // Proc of the IEEE Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2017: 2790-2798.
[22] MAO X J, SHEN C H, YANG Y B. Image Restoration Using very Deep Convolutional Encoder-Decoder Networks with Symmetric Skip Connections // Proc of the 30th International Conference on Neural Information Processing Systems. Cambridge, USA: MIT Press, 2016: 2810-2818.
[23] KIM J, LEE J K, LEE K M.Accurate Image Super-Resolution Using Very Deep Convolutional Networks // Proc of the IEEE Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2016: 1646-1654.
[24] ZHANG K, ZUO W M, CHEN Y J, et al. Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising. IEEE Transactions on Image Processing, 2017, 26(7): 3142-3155.
[25] TAI Y, YANG J, LIU X M, et al. MemNet: A Persistent Memory Network for Image Restoration // Proc of the IEEE International Conference on Computer Vision. Washington, USA: IEEE, 2017: 4549-4557.
[26] HUI Z, WANG X M, GAO X B.Fast and Accurate Single Image Super-Resolution via Information Distillation Network // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2018: 723-731.
[27] REN H Y, EL-KHAMY M, LEE J.Image Super Resolution Based on Fusing Multiple Convolution Neural Networks // Proc of the IEEE Conference on Computer Vision and Pattern Recognition Workshops. Washington, USA: IEEE, 2017: 1050-1057.
[28] CHOI J, KIM M.A Deep Convolutional Neural Network with Selection Units for Super-Resolution // Proc of the IEEE Conference on Computer Vision and Pattern Recognition Workshops. Washington, USA: IEEE, 2017: 1150-1156.
[29] AHN N, KANG B, SOHN K. Fast, Accurate, Lightweight Super-Resolution with Cascading Residual Network // Proc of the 15th European Conference on Computer Vision. Berlin, Germany: Springer, 2018: 256-272.
[30] KIM J, CHOI J, CHEON M, et al. MAMNet: Multi-path Adaptive Modulation Network for Image Super-Resolution. Neurocomputing, 2020, 402: 38-49.
[31] CHU X X, ZHANG B, MA H L, et al. Fast, Accurate and Lightweight Super-Resolution with Neural Architecture Search // Proc of the 25th International Conference on Pattern Recognition. Washington, USA: IEEE, 2021: 59-64.
[32] LI Z, WANG C F, SHI J.Lightweight Image Super-Resolution with Adaptive Weighted Learning Network. Computer Vision and Image Understanding, 2021, 211. DOI: 10.1016/j.cviu.2021.103254.
[33] LI Z, YANG J L, LIU Z, et al. Feedback Network for Image Super-Resolution // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2019: 3862-3871.
[34] HARIS M, SHAKHNAROVICH G, UKITA N.Deep Back-Projection Networks for Super-Resolution // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2018: 1664-1673.
[35] LI J C, FANG F M, MEI K F, et al. Multi-scale Residual Network for Image Super-Resolution // Proc of the 15th European Confe-rence on Computer Vision. Berlin, Germany: Springer, 2018: 527-542.