The application of convolutional neural networks for tomographic reconstruction of hyperspectral images

Wei Chih Huang; Mads Svanborg Peters; Mads Juul Ahlebæk; Mads Toudal Frandsen; René Lynge Eriksen; Bjarke Jørgensen

doi:10.1016/j.displa.2022.102218

The application of convolutional neural networks for tomographic reconstruction of hyperspectral images

Wei Chih Huang^*, Mads Svanborg Peters, Mads Juul Ahlebæk, Mads Toudal Frandsen, René Lynge Eriksen, Bjarke Jørgensen

^*Corresponding author for this work

Research output: Contribution to journal › Journal article › Research › peer-review

29 Downloads (Pure)

Abstract

A novel method, utilizing convolutional neural networks (CNNs), is proposed to reconstruct hyperspectral cubes from computed tomography imaging spectrometer (CTIS) images. Current reconstruction algorithms are usually subject to long reconstruction times and mediocre precision in cases of a large number of spectral channels. The constructed CNNs deliver higher precision and shorter reconstruction time than a sparse expectation maximization algorithm. In addition, the network can handle two different types of real-world images at the same time—specifically ColorChecker and carrot spectral images are considered. This work paves the way toward real-time reconstruction of hyperspectral cubes from CTIS images.

Original language	English
Article number	102218
Journal	Displays
Volume	74
ISSN	0141-9382
DOIs	https://doi.org/10.1016/j.displa.2022.102218
Publication status	Published - Sept 2022

Keywords

Computed tomography imaging spectrometer (CTIS) images
Convolutional neural networks
Hyperspectral cubes reconstruction

Documents & Links

10.1016/j.displa.2022.102218Licence: CC BY

Open access versionFinal published version, 6.79 MBLicence: CC BY

1 Finished

Food & Bio Cluster DK: Snapshot Hyperspectral imaging for food science across multiple scales
Frandsen, M. T., Eriksen, R. L., Huang, W. & Jørgensen, B.
01/05/2022 → 28/02/2023
Project: Research

Cite this

@article{b608e9b3d7dc4753b1ab48a5c1d65387,

title = "The application of convolutional neural networks for tomographic reconstruction of hyperspectral images",

abstract = "A novel method, utilizing convolutional neural networks (CNNs), is proposed to reconstruct hyperspectral cubes from computed tomography imaging spectrometer (CTIS) images. Current reconstruction algorithms are usually subject to long reconstruction times and mediocre precision in cases of a large number of spectral channels. The constructed CNNs deliver higher precision and shorter reconstruction time than a sparse expectation maximization algorithm. In addition, the network can handle two different types of real-world images at the same time—specifically ColorChecker and carrot spectral images are considered. This work paves the way toward real-time reconstruction of hyperspectral cubes from CTIS images.",

keywords = "Computed tomography imaging spectrometer (CTIS) images, Convolutional neural networks, Hyperspectral cubes reconstruction",

author = "Huang, {Wei Chih} and Peters, {Mads Svanborg} and Ahleb{\ae}k, {Mads Juul} and Frandsen, {Mads Toudal} and Eriksen, {Ren{\'e} Lynge} and Bjarke J{\o}rgensen",

note = "Funding Information: MTF and WCH acknowledge partial funding from the Independent Research Fund Denmark , grant number DFF 6108–00623 , The Villum Foundation and a CenSec grant funded by The Danish Ministry of Higher Education and Science (CenSec). MSP acknowledge partial funding from the Innovation Fund Denmark (IFD) under File No. 1044-00053B . This work was performed using the UCloud computing and storage resources, managed and supported by eScience center at SDU. Publisher Copyright: {\textcopyright} 2022 The Author(s)",

year = "2022",

month = sep,

doi = "10.1016/j.displa.2022.102218",

language = "English",

volume = "74",

journal = "Displays",

issn = "0141-9382",

publisher = "Elsevier",

}

TY - JOUR

T1 - The application of convolutional neural networks for tomographic reconstruction of hyperspectral images

AU - Huang, Wei Chih

AU - Peters, Mads Svanborg

AU - Ahlebæk, Mads Juul

AU - Frandsen, Mads Toudal

AU - Eriksen, René Lynge

AU - Jørgensen, Bjarke

N1 - Funding Information: MTF and WCH acknowledge partial funding from the Independent Research Fund Denmark , grant number DFF 6108–00623 , The Villum Foundation and a CenSec grant funded by The Danish Ministry of Higher Education and Science (CenSec). MSP acknowledge partial funding from the Innovation Fund Denmark (IFD) under File No. 1044-00053B . This work was performed using the UCloud computing and storage resources, managed and supported by eScience center at SDU. Publisher Copyright: © 2022 The Author(s)

PY - 2022/9

Y1 - 2022/9

N2 - A novel method, utilizing convolutional neural networks (CNNs), is proposed to reconstruct hyperspectral cubes from computed tomography imaging spectrometer (CTIS) images. Current reconstruction algorithms are usually subject to long reconstruction times and mediocre precision in cases of a large number of spectral channels. The constructed CNNs deliver higher precision and shorter reconstruction time than a sparse expectation maximization algorithm. In addition, the network can handle two different types of real-world images at the same time—specifically ColorChecker and carrot spectral images are considered. This work paves the way toward real-time reconstruction of hyperspectral cubes from CTIS images.

AB - A novel method, utilizing convolutional neural networks (CNNs), is proposed to reconstruct hyperspectral cubes from computed tomography imaging spectrometer (CTIS) images. Current reconstruction algorithms are usually subject to long reconstruction times and mediocre precision in cases of a large number of spectral channels. The constructed CNNs deliver higher precision and shorter reconstruction time than a sparse expectation maximization algorithm. In addition, the network can handle two different types of real-world images at the same time—specifically ColorChecker and carrot spectral images are considered. This work paves the way toward real-time reconstruction of hyperspectral cubes from CTIS images.

KW - Computed tomography imaging spectrometer (CTIS) images

KW - Convolutional neural networks

KW - Hyperspectral cubes reconstruction

U2 - 10.1016/j.displa.2022.102218

DO - 10.1016/j.displa.2022.102218

M3 - Journal article

SN - 0141-9382

VL - 74

JO - Displays

JF - Displays

M1 - 102218

ER -

The application of convolutional neural networks for tomographic reconstruction of hyperspectral images

Abstract

Keywords

Documents & Links

Fingerprint

Related projects

Food & Bio Cluster DK: Snapshot Hyperspectral imaging for food science across multiple scales

Cite this