3 minute read

扩大图像的分辨率,获得清晰画质;在医疗和卫星成像上有着重要应用;

super resolution

2019 年 超分辨率顶会论文数量真的是爆炸了; 好奇怪,这个方向综述论文出奇的多,别的方向都是总共不过 1-2 篇;

1 综述

  1. Super-resolution image reconstruction: A technical overview
    2003-05 paper

  2. A Survey on Various Single Image Super Resolution Techniques
    2012-12 paper

  3. Survey on Single image Super Resolution Techniques
    2013-04 paper

  4. Super-resolution: a comprehensive survey
    2014 paper

  5. A Survey on Various Techniques of Super Resolution Imaging
    2014-03 paper

  6. A Survey on Super-Resolution Methods for Image Reconstruction
    2014-03 paper

  7. A Survey of Single Image and Multi Image Super Resolution Techniques
    2014-11 paper

  8. A Short Survey of Image Super Resolution Algorithms
    2015 paper

  9. A Survey on Single Image Super Resolution Techniques
    2016 paper

  10. A survey of Super resolution Techniques
    2016-12 paper

  11. Deep Learning for Single Image Super-Resolution: A Brief Review
    2018-08-09 paper

  12. Deep Learning for Image Super-resolution: A Survey
    2019-02-16 paper | blog
    $\bullet \bullet$

  13. A Deep Journey into Super-resolution: A survey
    2019-04-16 paper

2 理论

3 基础研究

  1. Learning a Deep Convolutional Network for Image Super-Resolution
    2014 paper | project | blog |

  2. Image Super-Resolution Using Deep Convolutional Networks
    2014-12-31 paper | project | blog | blog,tensorflow

  3. To learn image super-resolution, use a GAN to learn how to do image degradation first
    ECCV 2018 2018-07-30 paper
    人工生成的低分辨率图像,和真实自然存在的图像并不相同;本文就对生成真实低分辩率图像做了相关研究;

4 注意力

  1. Hybrid Residual Attention Network for Single Image Super Resolution
    2019-07-11 paper

  2. Image Super-Resolution Using Very Deep Residual Channel Attention Networks
    ECCV 2018 2018-07-08 paper | pytorch
    深度残差注意力;

  3. Second-Order Attention Network for Single Image Super-Resolution
    CVPR 2019 2019 清华、鹏城实验室、香港理工大学、阿里达摩院 paper | PyTorch
    SAN 二阶注意力网络用于图像超分辨率

5 多尺度

  1. Deep Laplacian Pyramid Networks for Fast and Accurate Super-Resolution
    CVPR 2017 2017-04-12 paper

  2. Single Image Super-Resolution via Cascaded Multi-Scale Cross Network
    2018-02-24 paper

  3. Multi-scale Residual Network for Image Super-Resolution
    ECCV 2018 2018 paper | pytorch
    多尺度残差网络;讲述了经典超分辨率网络,都难以复现,扩展性有限;于是提出了新的结构;

  4. Multi-scale deep neural networks for real image super-resolution
    2019-04-24 paper | tensorflow-offical

6 级联

  1. Fast, Accurate, and Lightweight Super-Resolution with Cascading Residual Network
    ECCV 2018 2018-03-23 paper | pytorch
    级联残差下的快速、准确、轻量级的超分辨率网络;

7 GAN

  1. Image super-resolution through deep learning
    2014-12-31 paper | tensorflow

  2. Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network
    CVPR 2016 oral 2016-09-15 paper | torch

  3. EnhanceGAN

  4. ESRGAN: Enhanced Super-Resolution Generative Adversarial Networks
    ECCV 2018 2018-09-01 paper | pytorch-offical
    ESRGAN

8 其他

  1. Accelerating the Super-Resolution Convolutional Neural Network
    ECCV 2016 2016-08-01 paper

  2. Pixel Recursive Super Resolution
    2017-02-02 google paper

  3. Image Super-Resolution via Deep Recursive Residual Network
    CVPR 2017 2017 paper
    DRRN

  4. Deep Residual Network with Enhanced Upscaling Module for Super-Resolution
    CVPR 2018 2018 paper

  5. Deep Back-Projection Networks for Super-Resolution
    CVPR 2018 2018-03-07 paper | project | pytorch-offical

  6. SRFeat: Single Image Super-Resolution with Feature Discrimination
    ECCV 2018 2018 paper | project | tensorflow-offical
    SRFeat 具有特征识别的单个图像超分辨率;作者认为均方误差不足以表示特征图的真实特点;所以,在特征图中加入了对抗性损失;

  7. Analyzing Perception-Distortion Tradeoff Using Enhanced Perceptual Super-Resolution Network
    ECCV 2018 2018-11-01 paper | pytorch
    EPSR 在失真和质量上做权衡;

  8. Residual Networks for Light Field Image Super-Resolution
    CVPR 2019 2019 北京交通大学、北京航空航天大学 paper | pytorch
    残差网络用于光场图像超分辨率;

  9. Natural and Realistic Single Image Super-Resolution With Explicit Natural Manifold Discrimination
    CVPR 2019 2019 首尔国立大学 paper | tensorflow
    NatSR 自然、逼真的单图像超分辨率;

  10. ODE-Inspired Network Design for Single Image Super-Resolution
    CVPR 2019 2019 中科院、中科院大学、阿里巴巴 paper | pytorch
    OISR 单幅图像超分辨,常微分方程启发的网络设计

  11. Image Super-Resolution by Neural Texture Transfer
    CVPR 2019 2019-03-03 Adobe、田纳西大学 paper | project | tensorflow-offical
    SRNTT 神经纹理迁移的图像超分辨率;

  12. Meta-SR: A Magnification-Arbitrary Network for Super-Resolution
    CVPR 2019 2019-03-03 中国科技大学、中科院、旷视、清华 paper | pytorch-offical
    Meta-SR 任意缩放因子的超分辨率方法;

  13. Feedback Network for Image Super-Resolution
    CVPR 2019 2019-03-23 四川大学、加州大学圣巴巴拉分校、大不列颠哥伦比亚大学、韩国仁川国立大学 paper | pytorch-offical
    SRFBN 反馈网络用于图像超分辨;

  14. Deep Plug-And-Play Super-Resolution for Arbitrary Blur Kernels
    CVPR 2019 2019-03-29 哈尔滨工业大学、香港理工大学、鹏城实验室、阿里达摩院 paper | pytorch
    DPSR 能够应对任意模糊核的即插即用深度超分辨率;

  15. Blind Super-Resolution With Iterative Kernel Correction
    CVPR 2019 2019-04-06 香港中文大学、哈尔滨工业大学、中科院深圳先进技术研究院-商汤联合实验室 paper
    迭代模糊核校正的盲超分辨率;

  16. Camera Lens Super-Resolution
    CVPR 2019 2019-04-06 中国科技大学 paper | tensorflow-offical
    CameraSR 考虑真实成像环境分辨率和视野关系的镜头超分辨率;

  17. Towards Real Scene Super-Resolution With Raw Images
    CVPR 2019 2019-05-29 商汤 paper | project | code-offical
    Raw 图像的真实场景超分辨率,模拟真实成像过程生成训练数据,基于相机Raw数据进行超分辨率;

  18. Flickr1024: A Large-Scale Dataset for Stereo Image Super-Resolution
    ICCV 2019 workshop 2019-03-15 paper

9 应用

9.1 视频

9.2 3D

  1. 3D Appearance Super-Resolution With Deep Learning
    CVPR 2019 2019-06-03 瑞士苏黎世联邦理工学院,微软 paper | pytorch-offical
    3DASR 3D 表面超分辨;

9.3 双目

  1. Learning Parallax Attention for Stereo Image Super-Resolution
    CVPR 2019 2019-03-14 国防科技大学、盲信号处理重点实验室 paper | pytorch
    PASSRnet 双目超分辨率算法,提出并行注意力模型;

9.4 高光谱图像

  1. Hyperspectral Image Super-Resolution With Optimized RGB Guidance
    CVPR 2019 2019 北京理工大学、日本国立情报学研究所、格灵深瞳 paper
    高光谱图像超分辨率;

9.5 人脸

  1. Face Super-resolution Guided by Facial Component Heatmaps
    ECCV 2018 2018 paper
    CARN 针对人脸特性设计了特殊的网络,需要更少的训练样本;数据集用的 DIV2K;易复现;

TOP

附录

A 参考资料

  1. Intelligent Image Enhancement and Restoration - from Prior Driven Model to Advanced Deep Learning
  2. Hyperspectral-Image-Super-Resolution-Benchmark
  3. 深度学习方法的超分辨率
  4. 图像超分辨率重构
  5. Super resolution Survey

Comments