《计算机应用》唯一官方网站 ›› 2026, Vol. 46 ›› Issue (10): 3324-3331.DOI: 10.11772/j.issn.1001-9081.2025081031

• 多媒体计算与计算机仿真 • 上一篇    

混合退化估计与动态残差‒残差密集块的盲图像超分辨率重建方法

张鑫1, 伊华伟1(), 赵梦园1, 王艳飞1, 张林宸2   

  1. 1.辽宁工业大学 电子与信息工程学院,辽宁 锦州 121001
    2.北部湾大学 电子与信息工程学院,广西 钦州 535011
  • 收稿日期:2025-09-08 修回日期:2025-11-01 接受日期:2025-11-13 发布日期:2025-11-18 出版日期:2026-10-10
  • 通讯作者: 伊华伟
  • 作者简介:张鑫(2001—),男,辽宁铁岭人,硕士研究生,主要研究方向:深度学习、图像超分辨率重建
    赵梦园(1999—),女,山西运城人,硕士研究生,主要研究方向:图像处理、超分辨率重建
    王艳飞(1977—),男,辽宁锦州人,高级工程师,博士,主要研究方向:医学图像处理
    张林宸(2004—),男,辽宁沈阳人,主要研究方向:图像超分辨率重建。
  • 基金资助:
    国家自然科学基金资助项目(12371363);辽宁省教育厅基本研究项目(JYTMS20230860);辽宁省教育厅2024年高等学校基本研究项目(LJ212410154020);兴辽英才计划项目(XLYC2403125)

Blind image super-resolution reconstruction method by integrating degradation estimation and dynamic residual-in-residual dense blocks

Xin ZHANG1, Huawei YI1(), Mengyuan ZHAO1, Yanfei WANG1, Linchen ZHANG2   

  1. 1.College of Electronics and Information Engineering,Liaoning University of Technology,Jinzhou Liaoning 121001,China
    2.School of Electronic and Information Engineering,Beibu Gulf University,Qinzhou Guangxi 535011,China
  • Received:2025-09-08 Revised:2025-11-01 Accepted:2025-11-13 Online:2025-11-18 Published:2026-10-10
  • Contact: Huawei YI
  • About author:ZHANG Xin, born in 2001, M. S. candidate. His research interests include deep learning, image super-resolution reconstruction.
    ZHAO Mengyuan, born in 1999, M. S. candidate. Her research interests include image processing, super-resolution reconstruction.
    WANG Yanfei, born in 1977, Ph. D., senior engineer. His research interests include medical image processing.
    ZHANG Linchen, born in 2004. His research interests include image super-resolution reconstruction.
  • Supported by:
    National Natural Science Foundation of China(12371363);Basic Research Project of Educational Department of Liaoning Province(JYTMS20230860);Basic Research Project of Higher Education Institutions of Educational Department of Liaoning Province in 2024(LJ212410154020);Xingliao Talent Program(XLYC2403125)

摘要:

盲图像超分辨率重建致力于在真实场景下将低分辨率(LR)图像恢复成清晰的高分辨率(HR)图像。目前大多数退化估计网络不能较好地估计LR图像的退化信息,并且超分辨率网络中传统残差?残差密集块(RRDB)的泛化能力较差,无法适应复杂的退化类型。针对以上问题,提出混合退化估计与动态残差?残差密集块(DRRDB)的盲图像超分辨率重建方法(IDD-SR)。首先,提出一种混合退化估计网络,并将它用于精确估计图像退化信息;其次,提出一种DRRDB,以提高超分辨率重建网络的普适性;最后,通过端到端联合优化,实现对复杂退化信息的精准预测与图像的高质量超分辨率重建。在Set14、BSD100和DRealSR数据集上的实验结果表明,相较于PDM-SRGAN,IDD-SR的4倍超分辨率峰值信噪比(PSNR)值分别提升了0.90 dB、0.71 dB和0.97 dB,验证了它具有更优的细节恢复能力与伪影抑制效果。

关键词: 图像重建, 盲图像超分辨率, 生成对抗网络, 退化估计网络, 残差?残差密集块

Abstract:

Blind image super-resolution reconstruction focuses on recovering Low-Resolution (LR) images into clear High-Resolution (HR) images in real-world scenarios. Most existing degradation estimation networks cannot accurately estimate the degradation information of LR images, and the traditional Residual-in-Residual Dense Blocks (RRDBs) in super-resolution networks have poor generalization ability, which fail to adapt to complex degradation types. To address the above issues, a blind image Super-Resolution reconstruction method by Integrating Degradation estimation and Dynamic residual-in-residual dense blocks (IDD-SR) was proposed. First, an integrating degradation estimation network was proposed and used for accurate estimation of image degradation information. Then, a Dynamic Residual-in-Residual Dense Block (DRRDB) was proposed to improve the universality of the super-resolution reconstruction network. Finally, accurate prediction of complex degradation information and high-quality super-resolution image reconstruction were achieved through end-to-end joint optimization. Experimental results on the Set14, BSD100, and DRealSR datasets demonstrate that the Peak Signal-to-Noise Ratio (PSNR) values of 4× super-resolution of IDD-SR are improved by 0.90 dB, 0.71 dB, and 0.97 dB, respectively, compared with those of PDM-SRGAN. It can be seen that the proposed method has superior detail recovery capability and artifact suppression effect.

Key words: image reconstruction, blind image super-resolution, generative adversarial network, degradation estimation network, Residual-in-Residual Dense Block (RRDB)

中图分类号: