《计算机应用》唯一官方网站 ›› 2023, Vol. 43 ›› Issue (7): 2280-2287.DOI: 10.11772/j.issn.1001-9081.2022060877

• 多媒体计算与计算机仿真 • 上一篇    

融合迭代反馈与注意力机制的图像超分辨重建方法

梁敏(), 刘佳艺, 李杰   

  1. 山西财经大学 信息学院,太原 030006
  • 收稿日期:2022-06-16 修回日期:2022-09-06 接受日期:2022-09-08 发布日期:2022-10-18 出版日期:2023-07-10
  • 通讯作者: 梁敏
  • 作者简介:梁敏(1979—),女,山西忻州人,副教授,博士,CCF会员,主要研究方向:图像处理、模式识别;
    刘佳艺(1994—),女,安徽合肥人,硕士研究生,主要研究方向:深度学习、计算机视觉;
    李杰(1986—),男,四川邛崃人,副教授,博士,CCF会员,主要研究方向:计算机视觉、三维超分辨计算。
  • 基金资助:
    山西省高等学校哲学社会科学研究项目(2021W058);山西省研究生创新项目(2021SY533)

Image super-resolution reconstruction method based on iterative feedback and attention mechanism

Min LIANG(), Jiayi LIU, Jie LI   

  1. School of Information,Shanxi University of Finance and Economics,Taiyuan Shanxi 030006,China
  • Received:2022-06-16 Revised:2022-09-06 Accepted:2022-09-08 Online:2022-10-18 Published:2023-07-10
  • Contact: Min LIANG
  • About author:LIANG Min, born in 1979, Ph. D., associate professor. Her research interests include image processing, pattern recognition.
    LIU Jiayi, born in 1994, M. S. candidate. Her research interests include deep learning, computer vision.
    LI Jie, born in 1986, Ph. D., associate professor. His research interests include computer vision, three-dimensional super-resolution calculation.
  • Supported by:
    Program for the Philosophy and Social Sciences Research of Higher Learning Institutions of Shanxi(2021W058);Postgraduate Innovation Project of Shanxi Province(2021Y533)

摘要:

针对图像超分辨重建过程中原始高清图片与低质量图像之间缺乏依赖关系、深度网络中特征图信息不分主次重构导致的图像高频信息高精度重构困难的问题,提出一种融合迭代反馈与注意力机制的单幅图像超分辨重建方法。首先使用频率分解模块分别提取图像中的高、低频信息,并将二者分别处理,使网络重点关注提取出的高频细节部分,增强方法在图像细节上的复原能力;其次通过通道注意力机制将重建的重点放在有效特征所在的特征通道上,增强网络提取特征图信息的能力;然后采用迭代反馈的思想,在反复重建和比对过程中增加图像的还原程度;最后通过重建模块生成输出图像。在Set5、Set14、BSD100、Urban100和Manga109基准数据集上的2倍、4倍和8倍放大实验中,与主流超分辨率方法相比,所提方法表现出更优越的性能。在Manga109数据集的8倍放大实验中,相较于传统插值方法和基于卷积神经网络的图像超分辨率算法(SRCNN),所提方法的峰值信噪比(PSNR)均值分别提升了约3.01 dB和2.32 dB。实验结果表明:所提方法能够降低重建过程中出现的误差,并有效重建出更精细的高分辨率图像。

关键词: 深度学习, 单幅图像超分辨重建, 迭代反馈, 注意力机制, 频率分解

Abstract:

To address the difficulties in reconstructing high-frequency information in image super-resolution reconstruction due to the lack of dependency between low-resolution and high-resolution images and the lack of order during the reconstruction of feature map, a single-image super-resolution reconstruction method based on iterative feedback and attention mechanism was proposed. Firstly, high- and low-frequency information in the image was extracted respectively by using frequency decomposition block, and the two kinds of information was processed respectively, so that the network focused on the extracted high-frequency details to increase the restoration ability of the method on image details. Secondly, through the channel-wise attention mechanism, the reconstruction focus was put on the feature channels with effective features to improve the network ability of extracting the feature map information. Thirdly, the iterative feedback idea was adopted to increase quality of the restored image in the process of repeated comparison and reconstruction. Finally, the output image was generated through the reconstruction block. The proposed method shows better performance in comparison with mainstream super-resolution methods in the 2×, 4× and 8× experiments on Set5, Set14, BSD100, Urban100 and Manga109 benchmark datasets. In the 8× experiments on Manga109 dataset, the proposed method improves Peak Signal-to-Noise Ratio (PSNR) by about 3.01 dB and 2.32 dB averagely and respectively compared to the traditional interpolation method and the Super-Resolution Convolutional Neural Network (SRCNN). Experimental results show that the proposed method can reduce the errors in the reconstruction process and effectively reconstruct finer high-resolution images.

Key words: deep learning, single image super-resolution reconstruction, iterative feedback, attention mechanism, frequency decomposition

中图分类号: