A comparison of deep‐learning‐based inpainting techniques for experimental X‐ray scattering

The implementation is proposed of image inpainting techniques for the reconstruction of gaps in experimental X‐ray scattering data. The proposed methods use deep learning neural network architectures, such as convolutional autoencoders, tunable U‐Nets, partial convolution neural networks and mixed‐s...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Journal of applied crystallography Ročník 55; číslo 5; s. 1277 - 1288
Hlavní autoři: Chavez, Tanny, Roberts, Eric J., Zwart, Petrus H., Hexemer, Alexander
Médium: Journal Article
Jazyk:angličtina
Vydáno: 5 Abbey Square, Chester, Cheshire CH1 2HU, England International Union of Crystallography 01.10.2022
Blackwell Publishing Ltd
International Union of Crystallography (IUCr)
Témata:
Algorithms
Artificial neural networks
Computer architecture
Correlation coefficient
Correlation coefficients
Deep learning
Harmonic functions
image inpainting
Image reconstruction
Machine learning
MATHEMATICS AND COMPUTING
mixed-scale dense networks
Neural networks
Research Papers
Scattering
tunable U-Nets
X-ray scattering
ISSN:1600-5767, 0021-8898, 1600-5767
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:The implementation is proposed of image inpainting techniques for the reconstruction of gaps in experimental X‐ray scattering data. The proposed methods use deep learning neural network architectures, such as convolutional autoencoders, tunable U‐Nets, partial convolution neural networks and mixed‐scale dense networks, to reconstruct the missing information in experimental scattering images. In particular, the recovered pixel intensities are evaluated against their corresponding ground‐truth values using the mean absolute error and the correlation coefficient metrics. The results demonstrate that the proposed methods achieve better performance than traditional inpainting algorithms such as biharmonic functions. Overall, tunable U‐Net and mixed‐scale dense network architectures achieved the best reconstruction performance among all the tested algorithms, with correlation coefficient scores greater than 0.9980. A number of machine‐learning‐based algorithms are presented for the reconstruction of gaps in experimental X‐ray scattering images through inpainting approaches.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
content type line 23
AC02-05CH11231; 5R21GM129649-02
National Institutes of Health (NIH)
USDOE Office of Science (SC), Basic Energy Sciences (BES). Scientific User Facilities Division
ISSN:1600-5767
0021-8898
1600-5767
DOI:10.1107/S1600576722007105