欢迎访问《图学学报》 分享到:

图学学报 ›› 2023, Vol. 44 ›› Issue (1): 1-15.DOI: 10.11996/JG.j.2095-302X.2023010001

• 综述 • 上一篇    下一篇

基于深度学习的图像超分辨率研究综述

李洪安1(), 郑峭雪1, 陶若霖1, 张敏1, 李占利1(), 康宝生2   

  1. 1. 西安科技大学计算机科学与技术学院,陕西 西安 710054
    2. 西北大学信息科学与技术学院,陕西 西安 710127
  • 收稿日期:2022-05-18 修回日期:2022-06-13 出版日期:2023-10-31 发布日期:2023-02-16
  • 通讯作者: 李占利
  • 作者简介:李洪安(1978-),男,副教授,博士。主要研究方向为计算机图形学、可视媒体计算。E-mail:honganli@xust.edu.cn
  • 基金资助:
    陕西省自然科学基础研究计划项目(2023-JC-YB-517);陕西省自然科学基础研究计划项目(2022JM-508);陕西财经职业技术学院高层次人才引进项目(2022KY01)

Review of image super-resolutionbased on deep learning

LI Hong-an1(), ZHENG Qiao-xue1, TAO Ruo-lin1, ZHANG Min1, LI Zhan-li1(), KANG Bao-sheng2   

  1. 1. College of Computer Science and Technology, Xi'an University of Science and Technology, Xi?an Shaanxi 710054, China
    2. School of Information Science and Technology, Northwest University, Xi?an Shaanxi 710127, China
  • Received:2022-05-18 Revised:2022-06-13 Online:2023-10-31 Published:2023-02-16
  • Contact: LI Zhan-li
  • About author:LI Hong-an (1978-), associate professor, Ph.D. His main research interests cover computer graphics and visual media computing. E-mail:honganli@xust.edu.cn
  • Supported by:
    Natural Science Basic Research Plan of Shaanxi Province(2023-JC-YB-517);Natural Science Basic Research Plan of Shaanxi Province(2022JM-508);High-Level Talent Introduction Project of Shaanxi Technical College of Finance and Economics(2022KY01)

摘要:

超分辨率(SR)是一类重要的数字图像处理技术,其根据一个观测者得到的低分辨率(LR)图像重建并输出一个相应的高分辨率(HR)图像,从而提高现代数字图像的分辨率。SR在数字图像压缩与传输、医学成像、遥感成像、视频感知与监控等学科中的研究与应用价值巨大。随着深度学习的快速发展,结合最新的深度学习方法,可以为SR问题提供创新性的解决方案。首先回顾SR的背景意义、发展过程以及将深度学习应用于SR的技术价值。其次简要介绍传统SR算法的基本方法、分类和优缺点;按照不同的实现技术和网络类型对基于深度学习的SR方法进行了分类介绍,重点分析对比了卷积神经网络(CNN)、残差网络(ResNet)和生成对抗网络(GAN)在SR中的应用。然后介绍主要评价指标和解决策略,并对不同的SR算法在标准数据集中的性能表现进行对比。最后总结基于深度学习的SR算法,并对未来发展趋势进行展望。

关键词: 超分辨率, 深度学习, 评价指标, 退化模型, 数据集

Abstract:

Super-resolution (SR) is an essential technology in digital image processing that reconstructs and produces a matching high-resolution (HR) image based on the low-resolution (LR) image obtained by an observer, thereby improving the resolution of modern digital images. This technology is of significant research and practical value in fields such as digital image compression and transmission, medical imaging, remote sensing imaging, and video perception and monitoring. With the rapid growth of deep learning, novel solutions for SR challenges can be obtained by combining the latest deep learning algorithms. First, discussions were made on the background, development, and technological value of applying deep learning to SR. Second, a brief overview was provided concerning the fundamental methodology, categorization, and advantages and disadvantages of the classic SR methods. Deep learning-based SR methods were categorized and introduced based on distinct implementation strategies and network types. The application of convolutional neural networks (CNN), residual networks (ResNet), and generative adversarial networks (GAN) in SR was investigated and contrasted. The major evaluation indices and solution methodologies were then presented, and the performance of several SR methods on typical data sets was compared. Finally, the deep learning-based SR method was summarized, and the future development trend was forecasted.

Key words: super-resolution, deep learning, evaluation index, degradation model, datasets

中图分类号: