Deep Learning Techniques for Inverse Problems in Imaging

被引：374

作者：

Ongie, Gregory ^{[1
]}

Jalal, Ajil ^{[2
]}

Metzler, Christopher A. ^{[3
]}

Baraniuk, Richard G. ^{[4
]}

Dimakis, Alexandros G. ^{[2
]}

Willett, Rebecca ^{[1
,5
]}

机构：

[1] Univ Chicago, Dept Stat, Chicago, IL 60637 USA

[2] Univ Texas Austin, Dept Elect & Comp Engn, Austin, TX 78712 USA

[3] Stanford Univ, Dept Elect Engn, Stanford, CA 94305 USA

[4] Rice Univ, Dept Elect & Comp Engn, Houston, TX 77005 USA

[5] Univ Chicago, Dept Stat & Comp Sci, Chicago, IL 60637 USA

来源：

IEEE JOURNAL ON SELECTED AREAS IN INFORMATION THEORY | 2020年 / 1卷 / 01期

关键词：

Machine learning; deep neural networks; inverse problems; computational imaging; image restoration; image reconstruction; NEURAL-NETWORKS; DENSITY-ESTIMATION; X-RAY; RECONSTRUCTION; SCATTERING; MRI; SUPERRESOLUTION; MICROSCOPY; FRAMEWORK; ALGORITHM;

D O I：

10.1109/JSAIT.2020.2991563

中图分类号：

TP [自动化技术、计算机技术];

学科分类号：

0812 ;

摘要：

Recent work in machine learning shows that deep neural networks can be used to solve a wide variety of inverse problems arising in computational imaging. We explore the central prevailing themes of this emerging area and present a taxonomy that can be used to categorize different problems and reconstruction methods. Our taxonomy is organized along two central axes: (1) whether or not a forward model is known and to what extent it is used in training and testing, and (2) whether or not the learning is supervised or unsupervised, i.e., whether or not the training relies on access to matched ground truth image and measurement pairs. We also discuss the tradeoffs associated with these different reconstruction approaches, caveats and common failure modes, plus open problems and avenues for future work.

引用

页码：39 / 56

页数：18

共 187 条

[1]

Metzler CA, 2020, Arxiv, DOI arXiv:2002.05856

[2]

Metzler CA, 2020, Arxiv, DOI arXiv:1805.10531

[3]

Metzler CA, 2019, Arxiv, DOI arXiv:1908.00620

[4]

Adler J, 2018, Arxiv, DOI arXiv:1811.05910