Overview of Blind Deblurring Methods for Single Image

doi:10.3778/j.issn.1673-9418.2106100

Journal of Frontiers of Computer Science and Technology ›› 2022, Vol. 16 ›› Issue (3): 552-564.DOI: 10.3778/j.issn.1673-9418.2106100

• Surveys and Frontiers • Previous Articles Next Articles

Overview of Blind Deblurring Methods for Single Image

LIU Liping, SUN Jian⁺(), GAO Shiyan

School of Artificial Intelligence, North China University of Science and Technology, Tangshan, Hebei 063210, China

Received:2021-06-28 Revised:2021-08-26 Online:2022-03-01 Published:2021-09-01
About author:LIU Liping, born in 1977, Ph.D., professor. Her research interests include pattern recognition and intelligent system, mining engineering, etc.
SUN Jian, born in 1997, M.S. candidate. His research interests include image processing, machine vision and pattern recognition.
GAO Shiyan, born in 1997, M.S. candidate. Her research interests include image processing, machine vision and pattern recognition.
Supported by:
Science and Technology Program of Hebei Province(20327218D);Graduate Student Innovation Fund of North China University of Science and Technology(2019B28);Fundamental Research Funds for the Provincial Universities of Hebei(JYG2019004)

单图像盲去模糊方法概述

刘利平, 孙建⁺(), 高世妍

华北理工大学人工智能学院,河北唐山 063210

通讯作者: + E-mail: 1125439094@qq.com
作者简介:刘利平（1977—）,女,河北唐山人,博士在读,教授,主要研究方向为模式识别与智能系统、矿业工程等。
孙建（1997—）,男,河北衡水人,硕士研究生,主要研究方向为图像处理、机器视觉、模式识别。
高世妍（1997—）,女,河北唐山人,硕士研究生,主要研究方向为图像处理、机器视觉、模式识别。
基金资助:
河北省省级科技计划(20327218D);华北理工大学研究生创新项目(2019B28);河北省省属高校基本科研业务费研究项目(JYG2019004)

Abstract

Abstract:

Image deblurring has been a research hotspot in computer vision and image processing for a long time. The motion blur or focus blur image caused by camera jitter, object motion or defocus will seriously affect the use and follow-up processing of the image. The traditional blind deblurring methods make use of different causes of image motion blur, dividing motion blur into global motion blur and local motion blur. This paper summarizes the methods and research status of image blind deblurring in recent years. Then, on the basis of deep learning image deblurring methods, the image deblurring methods and research status are summarized. At the same time, the traditional blind deblurring methods and deep learning blind deblurring methods are classified and summarized, and the three forms of datasets needed before image deblurring and the quality evaluation criteria after image deblurring are summarized. Then, some of the traditional deblurring and deep learning deblurring methods are quantitatively and qualitatively analyzed and compared on the public deblurring dataset. Finally, the problems faced by the current image deblurring methods are analyzed, the research trend of image deblurring methods is prospected, and the main problems existing in single image blind deblurring are analyzed. The available solutions or ideas are explained one by one, which provides a theoretical basis for the follow-up research.

Key words: blind deblurring, motion blur, deep learning

摘要：

图像去模糊长期以来一直是计算机视觉和图像处理中的研究热点。由相机抖动、物体运动或失焦引起的运动模糊或焦点模糊图像会严重影响图像的使用和后续处理。传统的盲去模糊方法利用图像运动模糊产生的不同原因,可将运动模糊分为全局运动模糊和局部运动模糊。概述了近年来图像盲去模糊的方法和研究现状。在深度学习图像去模糊方法的基础上,总结了图像去模糊的方法及研究现状。同时,对传统的盲去模糊方法和深度学习的盲去模糊方法进行了分类总结,归纳了图像去模糊前所需数据集的三种构成方式和图像去模糊后的质量评价标准,然后在公共去模糊数据集上对部分传统去模糊和深度学习去模糊的方法进行了定量和定性的分析比较。最后,分析了当前图像去模糊方法面临的问题,展望了图像去模糊方法的研究趋势,并且对目前单图像盲去模糊存在的主要问题进行了拓展分析,逐一细化地说明了可用的解决方法或解决思路,为后续的研究提供了一定的理论依据。

关键词: 盲去模糊, 运动模糊, 深度学习

CLC Number:

TP391.4

LIU Liping, SUN Jian, GAO Shiyan. Overview of Blind Deblurring Methods for Single Image[J]. Journal of Frontiers of Computer Science and Technology, 2022, 16(3): 552-564.

刘利平, 孙建, 高世妍. 单图像盲去模糊方法概述[J]. 计算机科学与探索, 2022, 16(3): 552-564.

Figures/Tables 10

Table 1 Traditional blind deblurring methods

方法	适用场景	机制	优势	局限性
ECP^[12]	运动模糊	BCP和DCP相结合	同时考虑了暗通道和亮通道信息,无需任何复杂处理技术	易受低光照条件影响和图像明暗程度影响
PSF估计^[13]	散焦模糊运动模糊	以亚像素精度恢复空间变化的PSF	快速准确,仅使用单个图像以亚像素分辨率解决空间变化的PSF	只能求解单峰的核
FastGyro^[21]	运动模糊	基于惯性提高现有特征检测器和描述符是运动模糊的鲁棒性	增加了监测点的数量,提供了更高的重复性,提高了检测器定位精度	不适合强烈的运动模糊,去模糊效果受陀螺仪测量角速度高度影响
VEM^[22]	运动模糊	基于变分贝叶斯推理的边缘选择算法	易于实现,对变化内容稳定	可能会陷入一些远离真相的局部最小值
LMG^[23]	均匀模糊	基于半二次分裂策略和从粗到细的MAP框架	可处理各种特定场景	处理包含非高斯噪声图像时效率低下,需大量时间迭代更新变量
CME^[24]	运动模糊	使用边缘轮廓的思想来直接预测阶梯边缘	速度快、适用性强、计算复杂度降低	无法处理不同的大规模模糊
GCRF^[25]	均匀模糊	通过数据驱动方法有效地从训练集中学习好的核融合模型	更准确恢复底层内核产生更清晰的反卷积结果,减少振铃伪影	核融合框架完全依赖于单个方法的核估计结果,改进有限度
傅里叶突发积累^[26]	抖动模糊	图像在傅里叶域进行加权平均,加权取决于傅里叶谱的幅度	速度快,占用内存小,不会引入大多数反卷积算法中存在的典型振铃	优化慢
ADMM^[27]	抖动模糊	优化边缘检测器	速度快,感知质量更好,可集成到手机	无法捕捉到模糊程度为0的完美图像
RGTV^[28]	均匀模糊高斯模糊	交替求解骨架图像和模糊核	促进来自模糊观察的清晰图像的双峰权重分布,享有理想的频谱特性	对非均匀散焦模糊估计不佳
PLMG和低秩先验^[29]	运动模糊	高效的交替方向乘子法及半二次方分裂法	有效抑制潜在图像振铃效应并保留大多细节	易受模糊图像噪声影响
L₀稀疏先验方法^[30]	合成模糊图像真实模糊图像	对标签图像的像素分布和梯度分布特征进行分析	有效地抑制图像边缘的振铃效应,速度更快	迭代过程容易陷入局部最优问题
混合梯度稀疏先验^[31]	真实模糊图像	图像高阶梯度的稀疏性与低阶梯度相结合来构造混合梯度正则项	恢复出更锐利的边缘和更平滑的细节信息	存在一定程度的振铃效应
暗像素先验方法^[32]	真实模糊运动模糊	模糊图像的暗像素是非稀疏的	适用于大量图像,具有更好普适性	无法扩展到模糊视频的复原中

Table 1 Traditional blind deblurring methods

方法	适用场景	机制	优势	局限性
ECP^[12]	运动模糊	BCP和DCP相结合	同时考虑了暗通道和亮通道信息,无需任何复杂处理技术	易受低光照条件影响和图像明暗程度影响
PSF估计^[13]	散焦模糊运动模糊	以亚像素精度恢复空间变化的PSF	快速准确,仅使用单个图像以亚像素分辨率解决空间变化的PSF	只能求解单峰的核
FastGyro^[21]	运动模糊	基于惯性提高现有特征检测器和描述符是运动模糊的鲁棒性	增加了监测点的数量,提供了更高的重复性,提高了检测器定位精度	不适合强烈的运动模糊,去模糊效果受陀螺仪测量角速度高度影响
VEM^[22]	运动模糊	基于变分贝叶斯推理的边缘选择算法	易于实现,对变化内容稳定	可能会陷入一些远离真相的局部最小值
LMG^[23]	均匀模糊	基于半二次分裂策略和从粗到细的MAP框架	可处理各种特定场景	处理包含非高斯噪声图像时效率低下,需大量时间迭代更新变量
CME^[24]	运动模糊	使用边缘轮廓的思想来直接预测阶梯边缘	速度快、适用性强、计算复杂度降低	无法处理不同的大规模模糊
GCRF^[25]	均匀模糊	通过数据驱动方法有效地从训练集中学习好的核融合模型	更准确恢复底层内核产生更清晰的反卷积结果,减少振铃伪影	核融合框架完全依赖于单个方法的核估计结果,改进有限度
傅里叶突发积累^[26]	抖动模糊	图像在傅里叶域进行加权平均,加权取决于傅里叶谱的幅度	速度快,占用内存小,不会引入大多数反卷积算法中存在的典型振铃	优化慢
ADMM^[27]	抖动模糊	优化边缘检测器	速度快,感知质量更好,可集成到手机	无法捕捉到模糊程度为0的完美图像
RGTV^[28]	均匀模糊高斯模糊	交替求解骨架图像和模糊核	促进来自模糊观察的清晰图像的双峰权重分布,享有理想的频谱特性	对非均匀散焦模糊估计不佳
PLMG和低秩先验^[29]	运动模糊	高效的交替方向乘子法及半二次方分裂法	有效抑制潜在图像振铃效应并保留大多细节	易受模糊图像噪声影响
L₀稀疏先验方法^[30]	合成模糊图像真实模糊图像	对标签图像的像素分布和梯度分布特征进行分析	有效地抑制图像边缘的振铃效应,速度更快	迭代过程容易陷入局部最优问题
混合梯度稀疏先验^[31]	真实模糊图像	图像高阶梯度的稀疏性与低阶梯度相结合来构造混合梯度正则项	恢复出更锐利的边缘和更平滑的细节信息	存在一定程度的振铃效应
暗像素先验方法^[32]	真实模糊运动模糊	模糊图像的暗像素是非稀疏的	适用于大量图像,具有更好普适性	无法扩展到模糊视频的复原中

Table 2 Blind deblurring method based on deep learning

方法	适用场景	机制	优势	局限性
空间变体RNN^[37]	运动模糊动态场景模糊	通过无线脉冲响应模型制定去模糊过程	权重可以从另一个网络中学习,针对不同模糊学习不同权重	需同时涉及大区域和空间变化结构
SRN^[38]	运动模糊	新的多尺度循环网络结构	减少了可训练参数的数量,提高了训练效率	受限于固定数据集和训练时期
DMPHN^[45]	运动模糊	类似于空间金字塔匹配的端到端CNN分层模型	所需滤波器较小,可快速推理	需要较大的GPU内存
UID-GAN^[46]	特定区域图像模糊	用分离模型来分割模糊图像的内容和模糊特征	无需成对的训练图像,无监督学习	无法保留文本图像的一些细节
DPSR^[47]	LR模糊图像	设计了一个新的SISR退化模型	深度即插即用框架,可处理任意模糊核	对于大多数真实图像与退化模型不匹配变现不佳
BIE-RVD^[48]	运动模糊	基于端到端可微结构的时空视频自动编码器	网络快、更准确（特别是大模糊）	训练时执行任务复杂
UMSN^[49]	人脸等自然图像模糊	独立学习特定类别特征,结合特征对人脸图像去模糊	去模糊后的图像有更高的识别精度且同时保留人脸重要部分	对特征不明显模糊图像处理不佳
DDMS^[50]	运动模糊	构建一种具有滤波变换和特征调制能力的全卷积结构	完全省去了多尺度处理和大型过滤器,实时去模糊	不适合图像高频重要内容图像的恢复
Dr-Net^[51]	任意模糊	使用从数据中学习参数的深度网络,对图像先验和数据保真度近端算子建模	网络快,更好地处理异构模糊	卷积层最佳权重必须通过优化框架找到
DeepGyro^[52]	运动模糊	将陀螺仪测量值整合到卷积神经网络中	最大限度减少了补丁边缘附近伪影,实时处理极端和空间变化的运动模糊并且不会损坏清晰图像	易受自然图像的光源影响
MCGAN^[53]	面部和文本模糊	引入恢复细节的新训练损失与多类GAN结合	可从低分辨率模糊输入中恢复特定对象类别的图像细节	重建的人脸可能包含棋盘伪影,多类图像训练时很难为所有图像类学习一个统一的模型
MSLS^[54]	均匀模糊非均匀模糊	从单一的模糊观测中同时恢复潜在的锐化图像和模糊核	恢复图像中的伪影更少,运行速度快	要求存在具有轮廓的对象,其大小要远远大于模糊内核的大小
RSRN^[55]	毫米波辐射模糊图像	多级残差递归结构、多尺度循环连接	网络训练更稳定,更好保留细节信息	对其他模糊类型图像效果不佳
深度多级小波变换^[56]	运动模糊	多尺度扩张稠密块（MDDB）空间域重建模块（SDRM）	增大网络感受野,降低了映射复杂度	训练过程较复杂

Table 2 Blind deblurring method based on deep learning

方法	适用场景	机制	优势	局限性
空间变体RNN^[37]	运动模糊动态场景模糊	通过无线脉冲响应模型制定去模糊过程	权重可以从另一个网络中学习,针对不同模糊学习不同权重	需同时涉及大区域和空间变化结构
SRN^[38]	运动模糊	新的多尺度循环网络结构	减少了可训练参数的数量,提高了训练效率	受限于固定数据集和训练时期
DMPHN^[45]	运动模糊	类似于空间金字塔匹配的端到端CNN分层模型	所需滤波器较小,可快速推理	需要较大的GPU内存
UID-GAN^[46]	特定区域图像模糊	用分离模型来分割模糊图像的内容和模糊特征	无需成对的训练图像,无监督学习	无法保留文本图像的一些细节
DPSR^[47]	LR模糊图像	设计了一个新的SISR退化模型	深度即插即用框架,可处理任意模糊核	对于大多数真实图像与退化模型不匹配变现不佳
BIE-RVD^[48]	运动模糊	基于端到端可微结构的时空视频自动编码器	网络快、更准确（特别是大模糊）	训练时执行任务复杂
UMSN^[49]	人脸等自然图像模糊	独立学习特定类别特征,结合特征对人脸图像去模糊	去模糊后的图像有更高的识别精度且同时保留人脸重要部分	对特征不明显模糊图像处理不佳
DDMS^[50]	运动模糊	构建一种具有滤波变换和特征调制能力的全卷积结构	完全省去了多尺度处理和大型过滤器,实时去模糊	不适合图像高频重要内容图像的恢复
Dr-Net^[51]	任意模糊	使用从数据中学习参数的深度网络,对图像先验和数据保真度近端算子建模	网络快,更好地处理异构模糊	卷积层最佳权重必须通过优化框架找到
DeepGyro^[52]	运动模糊	将陀螺仪测量值整合到卷积神经网络中	最大限度减少了补丁边缘附近伪影,实时处理极端和空间变化的运动模糊并且不会损坏清晰图像	易受自然图像的光源影响
MCGAN^[53]	面部和文本模糊	引入恢复细节的新训练损失与多类GAN结合	可从低分辨率模糊输入中恢复特定对象类别的图像细节	重建的人脸可能包含棋盘伪影,多类图像训练时很难为所有图像类学习一个统一的模型
MSLS^[54]	均匀模糊非均匀模糊	从单一的模糊观测中同时恢复潜在的锐化图像和模糊核	恢复图像中的伪影更少,运行速度快	要求存在具有轮廓的对象,其大小要远远大于模糊内核的大小
RSRN^[55]	毫米波辐射模糊图像	多级残差递归结构、多尺度循环连接	网络训练更稳定,更好保留细节信息	对其他模糊类型图像效果不佳
深度多级小波变换^[56]	运动模糊	多尺度扩张稠密块（MDDB）空间域重建模块（SDRM）	增大网络感受野,降低了映射复杂度	训练过程较复杂

Fig.1 Results of different methods of deblurring pendulum image in Kohler dataset

Fig.2 Qualitative comparison of “face2” test images of Lai dataset

Table 3 PSNR and SSIM of different methods on GoPro dataset

方法	PSNR/dB	SSIM
Zhang et al（SVRNN）^[37]	29.19	0.930 6
Tao et al（SRN）^[38]	30.26	0.934 2
Kupyn et al（Deblur GAN）^[43]	28.70	0.958 0
Kupyn et al（DeblurGAN-v2）^[44]	29.55	0.934 5
Aljadaany et al（Dr-Net）^[51]	30.35	0.961 0
Sun et al（KCNN）^[58]	24.64	0.842 9
Nah et al（MSCNN）^[61]	29.08	0.913 5
Pan et al（暗通道先验）^[64]	23.52	0.833 6
Whyte et al（参数化几何模型）^[66]	24.53	0.845 8
Xu et al（ $L 0$ 稀疏表达式）^[68]	20.30	0.740 7
Kim et al（新的能量模型）^[69]	23.64	0.823 9
Liu et al（混合神经网络的低层视觉递归滤波器）^[70]	25.75	0.865 4
Gong et al（MBMF）^[71]	26.06	0.863 2

Table 3 PSNR and SSIM of different methods on GoPro dataset

方法	PSNR/dB	SSIM
Zhang et al（SVRNN）^[37]	29.19	0.930 6
Tao et al（SRN）^[38]	30.26	0.934 2
Kupyn et al（Deblur GAN）^[43]	28.70	0.958 0
Kupyn et al（DeblurGAN-v2）^[44]	29.55	0.934 5
Aljadaany et al（Dr-Net）^[51]	30.35	0.961 0
Sun et al（KCNN）^[58]	24.64	0.842 9
Nah et al（MSCNN）^[61]	29.08	0.913 5
Pan et al（暗通道先验）^[64]	23.52	0.833 6
Whyte et al（参数化几何模型）^[66]	24.53	0.845 8
Xu et al（ $L 0$ 稀疏表达式）^[68]	20.30	0.740 7
Kim et al（新的能量模型）^[69]	23.64	0.823 9
Liu et al（混合神经网络的低层视觉递归滤波器）^[70]	25.75	0.865 4
Gong et al（MBMF）^[71]	26.06	0.863 2

Table 4 PSNR and SSIM of different methods on Kohler dataset

方法	PSNR/dB	SSIM
Tao et al（SRN）^[38]	26.75	0.837 0
Kupyn et al（Deblur GAN）^[43]	26.10	0.816 0
Kupyn et al（DeblurGAN-v2）^[44]	26.72	0.836 0
Aljadaany et al（DR-Net）^[51]	27.20	0.865 0
Sun et al（KCNN）^[58]	25.22	0.773 5
Nah et al（MSCNN）^[61]	26.48	0.807 9
Whyte et al（参数化几何模型）^[66]	27.03	0.809 0
Xu et al（ $L 0$ 稀疏表达式）^[68]	27.47	0.811 0
Kim et al（新的能量模型）^[69]	24.68	0.793 7

Table 4 PSNR and SSIM of different methods on Kohler dataset

方法	PSNR/dB	SSIM
Tao et al（SRN）^[38]	26.75	0.837 0
Kupyn et al（Deblur GAN）^[43]	26.10	0.816 0
Kupyn et al（DeblurGAN-v2）^[44]	26.72	0.836 0
Aljadaany et al（DR-Net）^[51]	27.20	0.865 0
Sun et al（KCNN）^[58]	25.22	0.773 5
Nah et al（MSCNN）^[61]	26.48	0.807 9
Whyte et al（参数化几何模型）^[66]	27.03	0.809 0
Xu et al（ $L 0$ 稀疏表达式）^[68]	27.47	0.811 0
Kim et al（新的能量模型）^[69]	24.68	0.793 7

Table 5 PSNR and SSIM of different methods on Helen dataset

方法	PSNR/dB	SSIM
Kupyn et al^[44]	26.45	0.880
Yasarla et al^[49]	27.75	0.897
Nah et al^[61]	24.12	0.823
Cho et al^[63]	16.82	0.574
Pan et al^[64]	20.93	0.727
Krishnan et al^[67]	19.30	0.670
Xu et al^[68]	20.11	0.711
Shan et al^[72]	19.57	0.670
Zhong et al^[73]	16.41	0.614
Shen et al^[74]	25.58	0.861

Table 6 PSNR and SSIM of different methods on CelebA dataset

方法	PSNR/dB	SSIM
Kupyn et al^[44]	25.42	0.884
Yasarla et al^[49]	26.62	0.908
Nah et al^[61]	22.43	0.832
Cho et al^[63]	13.03	0.445
Pan et al^[64]	18.59	0.677
Krishnan et al^[67]	18.38	0.672
Xu et al^[68]	18.93	0.685
Shan et al^[72]	18.43	0.644
Zhong et al^[73]	17.26	0.695
Shen et al^[74]	24.34	0.860

Fig.3 Results of deblurring on real blurred image

Fig.4 Results of different methods on Starbucks blurred image processing

References 77

[1]	STOCKHAM T G, CANNON T M, INGEBRETSEN R B. Blind deconvolution through digital signal processing[J]. Proceedings of the IEEE, 1975, 63(4): 678-692. DOI URL
[2]	OPPENHEIM A, SCHAFER R, STOCKHAM T. Nonlinear filtering of multiplied and convolved signals[J]. IEEE Tran-sactions on Audio and Electroacoustics, 1968, 16(3): 437-466.
[3]	FERGUS R, SINGH B, HERTZMANN A, et al. Removing camera shake from a single photograph[J]. ACM Transac-tions on Graphics, 2006, 25(3): 787-794.
[4]	YOU Y L, KAVEH M. Blind image restoration by aniso-tropic regularization[J]. IEEE Transactions on Image Proce-ssing, 1999, 8(3): 396-407.
[5]	CHAN T F, WONG C K. Total variation blind deconvo-lution[J]. IEEE Transactions on Image Processing, 1998, 7(3): 370-375. DOI URL
[6]	KRISHNAN D, FERGUS R. Fast image deconvolution using hyper-Laplacian priors[C]// Proceedings of the 22nd Inter-national Conference on Neural Information Processing Sys-tems, Vancouver, Dec 7-10, 2009. Red Hook: Curran Asso-ciates, 2009: 1033-1041.
[7]	PAN J S, SU Z X. Fast ℓ⁰ -regularized kernel estimation for robust motion deblurring[J]. IEEE Signal Processing Let-ters, 2013, 20(9): 841-844.
[8]	PAN J S, LIN Z C, SU Z X, et al. Robust kernel estimation with outliers handling for image deblurring[C]// Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, Jun 27-30, 2016. Washington: IEEE Computer Society, 2016: 2800-2808.
[9]	PAN J S, HU Z, SU Z X, et al. L₀-regularized intensity and gradient prior for deblurring text images and beyond[J]. IEEE Transactions on Pattern Analysis and Machine Intel-ligence, 2016, 39(2): 342-355.
[10]	PAN J S, SUN D Q, PFISTER H, et al. Deblurring images via dark channel prior[J]. IEEE Transactions on Pattern Ana-lysis and Machine Intelligence, 2017, 40(10): 2315-2328.
[11]	HE K M, SUN J, TANG X O. Single image haze removal using dark channel prior[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2011, 33(12): 2341-2353. DOI URL
[12]	YAN Y Y, REN W Q, GUO Y F, et al. Image deblurring via extreme channels prior[C]// Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, Jul 21-26, 2017. Washington: IEEE Computer So-ciety, 2017: 6978-6986.
[13]	JOSHI N, SZELISKI R, KRIEGMAN D J. PSF estimation using sharp edge prediction[C]// Proceedings of the 2008 IEEE Conference on Computer Vision and Pattern Recog-nition, Anchorage, Jun 24-26, 2008. Washington: IEEE Com-puter Society, 2008: 1-8.
[14]	JIA J Y. Single image motion deblurring using transparency[C]// Proceedings of the 2007 IEEE Conference on Computer Vision and Pattern Recognition, Minneapolis, Jun 18-23, 2007. Washington: IEEE Computer Society, 2007: 1-8.
[15]	BAR L, SOCHEN N A, KIRYATI N. Restoration of images with piecewise space-variant blur[C]// LNCS 4485: Procee-dings of the 2007 International Conference on Scale Space and Variational Methods in Computer Vision, Ischia, May 30-Jun 2, 2007. Berlin, Heidelberg: Springer, 2007: 533-544.
[16]	SOREL M, FLUSSER J. Space-variant restoration of images degraded by camera motion blur[J]. IEEE Transactions on Image Processing, 2008, 17(2): 105-116. DOI URL
[17]	XU L, JIA J Y. Two-phase kernel estimation for robust mo-tion deblurring[C]// LNCS 6311: Proceedings of the 11th European Conference on Computer Vision, Heraklion, Sep 5-11, 2010. Berlin, Heidelberg: Springer, 2010: 157-170.
[18]	BAE S, DURAND F. Defocus magnification[J]. Computer Graphics Forum, 2007, 26(3): 571-579. DOI URL
[19]	DAI S Y, WU Y. Motion from blur[C]// Proceedings of the 2008 IEEE Conference on Computer Vision and Pattern Re-cognition, Anchorage, Jun 24-26, 2008. Washington: IEEE Computer Society, 2008: 1-8.
[20]	LEVIN A. Blind motion deblurring using image statistics[C]// Proceedings of the 20th Annual Conference on Neural Information Processing Systems, Vancouver, Dec 4-7, 2006. Cambridge: MIT Press, 2006: 841-848.
[21]	MUSTANIEMI J, KANNALA J, SÄRKKÄ S, et al. Fast motion deblurring for feature detection and matching using inertial measurements[C]// Proceedings of the 24th Interna-tional Conference on Pattern Recognition, Beijing, Aug 20-24, 2018. Washington: IEEE Computer Society, 2018: 3068-3073.
[22]	YANG L G, JI H. A variational EM framework with adap-tive edge selection for blind motion deblurring[C]// Procee-dings of the 2019 IEEE Conference on Computer Vision and Pattern Recognition, Long Beach, Jun 16-20, 2019. Pis-cataway: IEEE, 2019: 10167-10176.
[23]	CHEN L, FANG F M, WANG T T, et al. Blind image deb-lurring with local maximum gradient prior[C]// Proceedings of the 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Long Beach, Jun 16-20, 2019. Pisca-taway: IEEE, 2019: 1742-1750.
[24]	VASU S, RAJAGOPALAN A N. From local to global: edge profiles to camera motion in blurred images[C]// Procee-dings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, Jul 21-26, 2017. Wa-shington: IEEE Computer Society, 2017: 4447-4456.
[25]	MAI L, LIU F. Kernel fusion for better image deblurring[C]// Proceedings of the 2015 IEEE Conference on Computer Vision and Pattern Recognition, Boston, Jun 7-12, 2015. Piscataway: IEEE, 2015: 371-380.
[26]	DELBRACIO M, SAPIRO G. Burst deblurring: removing camera shake through Fourier burst accumulation[C]// Pro-ceedings of the 2015 IEEE Conference on Computer Vision and Pattern Recognition, Boston, Jun 7-12, 2015. Piscata-way: IEEE, 2015: 2385-2393.
[27]	PENDYALA S, RAMESHA P, BNS A V, et al. Blur de-tection and fast blind image deblurring[C]// Proceedings of the 2015 Annual IEEE India Conference, New Delhi, Dec 17-20, 2015. Piscataway: IEEE, 2015: 1-4.
[28]	BAI Y C, CHEUNG G, LIU X M, et al. Graph-based blind image deblurring from a single photograph[J]. IEEE Tran-sactions on Image Processing, 2018, 28(3): 1404-1418.
[29]	周志豪, 张玉龙, 唐启凡, 等. 多尺度低秩图像盲去模糊方法[J]. 西安交通大学学报, 2021, 55(9): 168-177.
	ZHOU Z H, ZHANG Y L, TANG Q F, et al. Multi-scale low-rank blind image deblurring method[J]. Journal of Xi’an Jiao-tong University, 2021, 55(9): 168-177.
[30]	柳宁, 赵焕明, 李德平, 等. 基于L0稀疏先验的运动模糊标签图像盲复原[J]. 华南理工大学学报(自然科学版), 2021, 49(3): 8-16.
	LIU N, ZHAO H M, LI D P, et al. Blind restoration of motion blur label image based on L0 sparse priors[J]. Journal of South China University of Technology (Natural Science Edition), 2021, 49(3): 8-16.
[31]	徐宁珊, 王琛, 任国强, 等. 混合梯度稀疏先验约束下的图像盲复原[J]. 光电工程, 2021, 48(6): 58-69.
	XU N S, WANG C, REN G Q, et al. Blind image resto-ration method regularized by hybrid gradient sparse prior[J]. Opto-Electronic Engineering, 2021, 48(6): 58-69.
[32]	涂春梅, 陈国彬, 刘超. 暗像素先验的模糊图像盲复原方法[J]. 计算机工程与应用, 2020, 56(10): 213-219.
	TU C M, CHEN G B, LIU C. Dark-pixel-prior blind deblur-ring method[J]. Computer Engineering and Applications, 2020, 56(10): 213-219.
[33]	CRONJE J. Deep convolutional neural networks for dense non-uniform motion deblurring[C]// Proceedings of the 2015 International Conference on Image and Vision Computing, New Zealand, Nov 23-24, 2015. Piscataway: IEEE, 2015: 1-5.
[34]	XU X Y, PAN J S, ZHANG Y J, et al. Motion blur kernel estimation via deep learning[J]. IEEE Transactions on Ima-ge Processing, 2018, 27(1): 194-205.
[35]	HRADIŠ M, KOTERA J, ZEMCÍK P, et al. Convolutional neural networks for direct text deblurring[C]// Proceedings of the British Machine Vision Conference 2015, Swansea, Sep 7-10, 2015. Durham: BMVA Press, 2015: 1-13.
[36]	SCHULER C J, HIRSCH M, HARMELING S, et al. Lear-ning to deblur[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 38(7): 1439-1451. DOI URL
[37]	ZHANG J W, PAN J S, REN J S J, et al. Dynamic scene deblurring using spatially variant recurrent neural networks[C]// Proceedings of the 2018 IEEE Conference on Com-puter Vision and Pattern Recognition, Salt Lake City, Jun 18-23, 2018. Piscataway: IEEE, 2018: 2521-2529.
[38]	TAO X, GAO H Y, SHEN X Y, et al. Scale-recurrent net-work for deep image deblurring[C]// Proceedings of the 2018 IEEE Conference on Computer Vision and Pattern Recog-nition, Salt Lake City, Jun 18-22, 2018. Piscataway: IEEE, 2018: 8174-8182.
[39]	GOODFELLOW I J, POUGET-ABADIE J, MIRZA M, et al. Generative adversarial networks[J]. arXiv:1406.2661, 2014.
[40]	LEDIG C, THEIS L, HUSZÁR F, et al. Photo-realistic single image super-resolution using a generative adversarial net-work[C]// Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, Jul 21-26, 2017. Washington: IEEE Computer Society, 2017: 4681-4690.
[41]	CHEN Y, WU F G, ZHAO J S. Motion deblurring via using generative adversarial networks for space-based imaging[C]// Proceedings of the 16th IEEE International Confe-rence on Software Engineering Research, Management and Applications, Kunming, Jun 13-15, 2018. Washington: IEEE Computer Society, 2018: 37-41.
[42]	GULRAJANI I, AHMED F, ARJOVSKY M, et al. Improved training of Wasserstein GANs[J]. arXiv:1704.00028, 2017.
[43]	KUPYN O, BUDZAN V, MYKHAILYCH M, et al. Deblur-GAN: blind motion deblurring using conditional adver-sarial networks[C]// Proceedings of the 2018 IEEE Confe-rence on Computer Vision and Pattern Recognition, Salt Lake City, Jun 18-22, 2018. Washington: IEEE Computer Society, 2018: 8183-8192.
[44]	KUPYN O, MARTYNIUK T, WU J R, et al. DeblurGAN-v2: deblurring (orders-of-magnitude) faster and better[C]// Proceedings of the 2019 IEEE/CVF International Conferen-ce on Computer Vision, Seoul, Oct 27-Nov 2, 2019. Piscata-way: IEEE, 2019: 8878-8887.
[45]	ZHANG H G, DAI Y C, LI H D, et al. Deep stacked hie-rarchical multi-patch network for image deblurring[C]// Pro-ceedings of the 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Long Beach, Jun 15-20, 2019. Piscataway: IEEE, 2019: 5978-5986.
[46]	LU B Y, CHEN J C, CHELLAPPA R. UID-GAN: unsuper-vised image deblurring via disentangled representations[J]. IEEE Transactions on Biometrics, Behavior, and Identity Science, 2020, 2(1): 26-39. DOI URL
[47]	ZHANG K, ZUO W M, ZHANG L. Deep plug-and-play super-resolution for arbitrary blur kernels[C]// Proceedings of the 2019 IEEE/CVF Conference on Computer Vision and Pat-tern Recognition, Long Beach, Jun 16-20, 2019. Piscataway:IEEE, 2019: 1671-1681.
[48]	PUROHIT K, SHAH A B, RAJAGOPALAN A N. Bringing alive blurred moments[C]// Proceedings of the 2019 IEEE Conference on Computer Vision and Pattern Recognition, Long Beach, Jun 16-20, 2019. Piscataway: IEEE, 2019: 6830-6839.
[49]	YASARLA R, PERAZZI F, PATEL V M. Deblurring face images using uncertainty guided multi-stream semantic net-works[J]. IEEE Transactions on Image Processing, 2020, 29: 6251-6263. DOI URL
[50]	PUROHIT K, RAJAGOPALAN A N. Region-adaptive dense network for efficient motion deblurring[C]// Proceedings of the 34th AAAI Conference on Artificial Intelligence, the 32nd Innovative Applications of Artificial Intelligence Con-ference, the 10th AAAI Symposium on Educational Advan-ces in Artificial Intelligence, New York, Feb 7-12, 2020. Menlo Park: AAAI, 2020: 11882-11889.
[51]	ALJADAANY R, PAL D K, SAVVIDES M. Douglas-Rachford networks: learning both the image prior and data fidelity terms for blind image deconvolution[C]// Procee-dings of the 2019 IEEE Conference on Computer Vision and Pattern Recognition, Long Beach, Jun 16-20, 2019. Pis-cataway: IEEE, 2019: 10235-10244.
[52]	MUSTANIEMI J, KANNALA J, SÄRKKÄ S, et al. Gyroscope-aided motion deblurring with deep networks[C]// Proceedings of the 2019 IEEE Winter Conference on Applica-tions of Computer Vision, Waikoloa Village, Jan 7-11, 2019. Piscataway: IEEE, 2019: 1914-1922.
[53]	XU X Y, SUN D Q, PAN J S, et al. Learning to super-resolve blurry face and text images[C]// Proceedings of the 2017 IEEE International Conference on Computer Vision, Venice, Oct 22-29, 2017. Washington: IEEE Computer So-ciety, 2017: 251-260.
[54]	BAI Y C, JIA H Z, JIANG M, et al. Single-image blind deb-lurring using multi-scale latent structure prior[J]. IEEE Tran-sactions on Circuits and Systems for Video Technology, 2019, 30(7): 2033-2045.
[55]	徐国豪, 刘媛媛, 朱路. 基于残差递归网络的毫米波辐射图像去模糊[J/OL]. 激光与光电子学进展 [2021-08-06]. http://kns.cnki.net/kcms/detail/31.1690.TN.20210804.1202.038.html.
	XU G H, LIU Y Y, ZHU L. Millimeter wave radiation ima-ge deblurring based on residual recursive network[J/OL]. Progress in Laser and Optoelectronics [2021-08-06]. http://kns.cnki.net/kcms/detail/31.1690.TN.20210804.1202.038.html.
[56]	陈书贞, 曹世鹏, 崔美玥, 等. 基于深度多级小波变换的图像盲去模糊算法[J]. 电子与信息学报, 2021, 43(1): 154-161.
	CHEN S Z, CAO S P, CUI M Y, et al. Image blind deblur-ring algorithm based on deep multi-level wavelet transform[J]. Journal of Electronics & Information Technology, 2021, 43(1): 154-161.
[57]	LEVIN A, WEISS Y, DURAND F, et al. Understanding and evaluating blind deconvolution algorithms[C]// Proceedings of the 2009 IEEE Conference on Computer Vision and Pattern Recognition, Miami, Jun 20-25, 2009. Washington: IEEE Com-puter Society, 2009: 1964-1971.
[58]	SUN L B, CHO S, WANG J, et al. Edge-based blur kernel estimation using patch priors[C]// Proceedings of the 2013 IEEE International Conference on Computational Photo-graphy, Cambridge, Apr 19-21, 2013. Washington: IEEE Com-puter Society, 2013: 1-8.
[59]	LAI W S, HUANG J B, HU Z, et al. A comparative study for single image blind deblurring[C]// Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Reco-gnition, Las Vegas, Jun 27-30, 2016. Washington: IEEE Com-puter Society, 2016: 1701-1709.
[60]	KÖHLER R, HIRSCH M, MOHLER B J, et al. Recording and playback of camera shake: benchmarking blind decon-volution with a real-world database[C]// LNCS 7578: Pro-ceedings of the 12th European Conference on Computer Vision, Florence, Oct 7-13, 2012. Berlin, Heidelberg: Sprin-ger, 2012: 27-40.
[61]	NAH S, KIM T H, LEE K M. Deep multi-scale convolu-tional neural network for dynamic scene deblurring[C]// Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, Jul 21-26, 2017. Washington: IEEE Computer Society, 2017: 257-265.
[62]	WANG Z, BOVIK A C, SHEIKH H R, et al. Image quality assessment: from error visibility to structural similarity[J]. IEEE Transactions on Image Processing, 2004, 13(4): 600-612. DOI URL
[63]	CHO S, LEE S. Fast motion deblurring[J]. ACM Transac-tions on Graphics, 2009, 28(5): 145.
[64]	PAN J S, SUN D Q, PFISTER H, et al. Blind image deblur-ring using dark channel prior[C]// Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recogni-tion, Las Vegas, Jun 27-30, 2016. Washington: IEEE Com-puter Society, 2016: 1628-1636.
[65]	LI L, PAN J S, LAI W S, et al. Learning a discriminative prior for blind image deblurring[C]// Proceedings of the 2018 IEEE Conference on Computer Vision and Pattern Recog-nition, Salt Lake City, Jun 18-23, 2018. Washington: IEEE Computer Society, 2018: 6616-6625.
[66]	WHYTE O, SIVIC J, ZISSERMAN A, et al. Non-uniform deblurring for shaken images[J]. International Journal of Com-puter Vision, 2012, 98(2): 168-186.
[67]	KRISHNAN D, TAY T, FERGUS R. Blind deconvolution using a normalized sparsity measure[C]// Proceedings of the 24th IEEE Conference on Computer Vision and Pattern Reco-gnition, Colorado Springs, Jun 20-25, 2011. Washington: IEEE Computer Society, 2011: 233-240.
[68]	XU L, ZHENG S C, JIA J Y. Unnatural L0 sparse represen-tation for natural image deblurring[C]// Proceedings of the 2013 IEEE Conference on Computer Vision and Pattern Re-cognition, Portland, Jun 23-28, 2013. Washington: IEEE Com-puter Society, 2013: 1107-1114.
[69]	KIM T Y, LEE K M. Segmentation-free dynamic scene deb-lurring[C]// Proceedings of the 2014 IEEE Conference on Computer Vision and Pattern Recognition, Columbus, Jun 23-28, 2014. Washington: IEEE Computer Society, 2014: 2766-2773.
[70]	LIU S F, PAN J S, YANG M H. Learning recursive filters for low-level vision via a hybrid neural network[C]// LNCS 9908: Proceedings of the 14th European Conference on Com-puter Vision, Amsterdam, Oct 11-14, 2016. Cham: Springer, 2016: 560-576.
[71]	GONG D, YANG J, LIU L Q, et al. From motion blur to mo-tion flow: a deep learning solution for removing hetero-geneous motion blur[C]// Proceedings of the 2017 IEEE Con-ference on Computer Vision and Pattern Recognition, Hono-lulu, Jul 21-26, 2017. Washington: IEEE Computer Society, 2017: 3806-3815.
[72]	SHAN Q, JIA J Y, AGARWALA A. High-quality motion deblurring from a single image[J]. ACM Transactions on Graphics, 2008, 27(3): 73.
[73]	ZHONG L, CHO S, METAXAS D N, et al. Handling noise in single image deblurring using directional filters[C]// Pro-ceedings of the 2013 IEEE Conference on Computer Vision and Pattern Recognition, Portland, Jun 23-28, 2013. Washin-gton: IEEE Computer Society, 2013: 612-619.
[74]	SHEN Z Y, LAI W S, XU T F, et al. Deep semantic face deblurring[C]// Proceedings of the 2018 IEEE Conference on Computer Vision and Pattern Recognition, Salt Lake City, Jun 18-22, 2018. Washington: IEEE Computer Society, 2018: 8260-8269.
[75]	PAN J S, HU Z, SU Z X, et al. Deblurring text images via L0-regularized intensity and gradient prior[C]// Proceedings of the 2014 IEEE Conference on Computer Vision and Pattern Recognition, Columbus, Jun 23-28, 2014. Washing-ton: IEEE Computer Society, 2014: 2901-2908.
[76]	MADAM N T, KUMAR S, RAJAGOPALAN A N. Unsu-pervised class-specific deblurring[C]// LNCS 11214: Procee-dings of the 15th European Conference on Computer Vi-sion, Munich, Sep 8-14, 2018. Cham: Springer, 2018: 358-374.
[77]	LU B Y, CHEN J C, CHELLAPPA R. Unsupervised domain-specific deblurring via disentangled representations[C]// Pro-ceedings of the 2019 IEEE Conference on Computer Vision and Pattern Recognition, Long Beach, Jun 16-20, 2019. Pis-cataway: IEEE, 2019: 10225-10234.

Overview of Blind Deblurring Methods for Single Image

单图像盲去模糊方法概述

RichHTML

PDF

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 10

References 77

Related Articles 15

Recommended Articles

Metrics

[1]	AN Fengping, LI Xiaowei, CAO Xiang. Medical Image Classification Algorithm Based on Weight Initialization-Sliding Window CNN [J]. Journal of Frontiers of Computer Science and Technology, 2022, 16(8): 1885-1897.
[2]	ZENG Fanzhi, XU Luqian, ZHOU Yan, ZHOU Yuexia, LIAO Junwei. Review of Knowledge Tracing Model for Intelligent Education [J]. Journal of Frontiers of Computer Science and Technology, 2022, 16(8): 1742-1763.
[3]	LIU Yi, LI Mengmeng, ZHENG Qibin, QIN Wei, REN Xiaoguang. Survey on Video Object Tracking Algorithms [J]. Journal of Frontiers of Computer Science and Technology, 2022, 16(7): 1504-1515.
[4]	ZHAO Xiaoming, YANG Yijiao, ZHANG Shiqing. Survey of Deep Learning Based Multimodal Emotion Recognition [J]. Journal of Frontiers of Computer Science and Technology, 2022, 16(7): 1479-1503.
[5]	XIA Hongbin, XIAO Yifei, LIU Yuan. Long Text Generation Adversarial Network Model with Self-Attention Mechanism [J]. Journal of Frontiers of Computer Science and Technology, 2022, 16(7): 1603-1610.
[6]	SUN Fangwei, LI Chengyang, XIE Yongqiang, LI Zhongbo, YANG Caidong, QI Jin. Review of Deep Learning Applied to Occluded Object Detection [J]. Journal of Frontiers of Computer Science and Technology, 2022, 16(6): 1243-1259.
[7]	LIU Yafen, ZHENG Yifeng, JIANG Lingyi, LI Guohe, ZHANG Wenjie. Survey on Pseudo-Labeling Methods in Deep Semi-supervised Learning [J]. Journal of Frontiers of Computer Science and Technology, 2022, 16(6): 1279-1290.
[8]	CHENG Weiyue, ZHANG Xueqin, LIN Kezheng, LI Ao. Deep Convolutional Neural Network Algorithm Fusing Global and Local Features [J]. Journal of Frontiers of Computer Science and Technology, 2022, 16(5): 1146-1154.
[9]	ZHONG Mengyuan, JIANG Lin. Review of Super-Resolution Image Reconstruction Algorithms [J]. Journal of Frontiers of Computer Science and Technology, 2022, 16(5): 972-990.
[10]	XU Jia, WEI Tingting, YU Ge, HUANG Xinyue, LYU Pin. Review of Question Difficulty Evaluation Approaches [J]. Journal of Frontiers of Computer Science and Technology, 2022, 16(4): 734-759.
[11]	PEI Lishen, ZHAO Xuezhuan. Survey of Collective Activity Recognition Based on Deep Learning [J]. Journal of Frontiers of Computer Science and Technology, 2022, 16(4): 775-790.
[12]	ZHU Weijie, CHEN Ying. Micro-expression Recognition Convolutional Network for Dual-stream Temporal-Domain Information Interaction [J]. Journal of Frontiers of Computer Science and Technology, 2022, 16(4): 950-958.
[13]	JIANG Yi, XU Jiajie, LIU Xu, ZHU Junwu. Research on Edge-Guided Image Repair Algorithm [J]. Journal of Frontiers of Computer Science and Technology, 2022, 16(3): 669-682.
[14]	ZHANG Quangui, HU Jiayan, WANG Li. One Class Collaborative Filtering Recommendation Algorithm Coupled with User Common Characteristics [J]. Journal of Frontiers of Computer Science and Technology, 2022, 16(3): 637-648.
[15]	WU Kaijun, HUANG Tao, WANG Dicong, BAI Chenshuai, TAO Xiaomiao. Research Progress of Video Anomaly Detection Technology [J]. Journal of Frontiers of Computer Science and Technology, 2022, 16(3): 529-540.