Dense U-Net for Limited Angle Tomography of Sound Pressure Fields

Dense U-Net for Limited Angle Tomography of Sound Pressure Fields

Tomographic reconstruction allows for the recovery of 3D information from 2D projection data. This commonly requires a full angular scan of the specimen. Angular restrictions that exist, especially in technical processes, result in reconstruction artifacts and unknown systematic measurement errors....

Full description

Bibliographic Details
Main Authors: Oliver Rothkamm, Johannes Gürtler, Jürgen Czarske, Robert Kuschmierz
Format: Article
Language:English
Published: MDPI AG 2021-05-01
Series:Applied Sciences
Subjects:
bias-flow liner
tomography
highspeed camera
volumetric sound pressure
dense U-Net
deep learning
Online Access:https://www.mdpi.com/2076-3417/11/10/4570
id doaj-77687ebde3bc4dc4869183c44e7e4de7
record_format Article
spelling doaj-77687ebde3bc4dc4869183c44e7e4de72021-06-01T00:16:04ZengMDPI AGApplied Sciences2076-34172021-05-01114570457010.3390/app11104570Dense U-Net for Limited Angle Tomography of Sound Pressure FieldsOliver Rothkamm0Johannes Gürtler1Jürgen Czarske2Robert Kuschmierz3Laboratory of Measurement and Sensor System Technique, Faculty of Electrical and Computer Engineering, TU Dresden, Helmholtzstraße 18, 01069 Dresden, GermanyLaboratory of Measurement and Sensor System Technique, Faculty of Electrical and Computer Engineering, TU Dresden, Helmholtzstraße 18, 01069 Dresden, GermanyLaboratory of Measurement and Sensor System Technique, Faculty of Electrical and Computer Engineering, TU Dresden, Helmholtzstraße 18, 01069 Dresden, GermanyLaboratory of Measurement and Sensor System Technique, Faculty of Electrical and Computer Engineering, TU Dresden, Helmholtzstraße 18, 01069 Dresden, GermanyTomographic reconstruction allows for the recovery of 3D information from 2D projection data. This commonly requires a full angular scan of the specimen. Angular restrictions that exist, especially in technical processes, result in reconstruction artifacts and unknown systematic measurement errors. We investigate the use of neural networks for extrapolating the missing projection data from holographic sound pressure measurements. A bias flow liner was studied for active sound dampening in aviation. We employed a dense U-Net trained on synthetic data and compared reconstructions of simulated and measured data with and without extrapolation. In both cases, the neural network based approach decreases the mean and maximum measurement deviations by a factor of two. These findings can enable quantitative measurements in other applications suffering from limited angular access as well.https://www.mdpi.com/2076-3417/11/10/4570bias-flow linertomographyhighspeed cameravolumetric sound pressuredense U-Netdeep learning
collection DOAJ
language English
format Article
sources DOAJ
author Oliver Rothkamm
Johannes Gürtler
Jürgen Czarske
Robert Kuschmierz
spellingShingle Oliver Rothkamm
Johannes Gürtler
Jürgen Czarske
Robert Kuschmierz
Dense U-Net for Limited Angle Tomography of Sound Pressure Fields
Applied Sciences
bias-flow liner
tomography
highspeed camera
volumetric sound pressure
dense U-Net
deep learning
author_facet Oliver Rothkamm
Johannes Gürtler
Jürgen Czarske
Robert Kuschmierz
author_sort Oliver Rothkamm
title Dense U-Net for Limited Angle Tomography of Sound Pressure Fields
title_short Dense U-Net for Limited Angle Tomography of Sound Pressure Fields
title_full Dense U-Net for Limited Angle Tomography of Sound Pressure Fields
title_fullStr Dense U-Net for Limited Angle Tomography of Sound Pressure Fields
title_full_unstemmed Dense U-Net for Limited Angle Tomography of Sound Pressure Fields
title_sort dense u-net for limited angle tomography of sound pressure fields
publisher MDPI AG
series Applied Sciences
issn 2076-3417
publishDate 2021-05-01
description Tomographic reconstruction allows for the recovery of 3D information from 2D projection data. This commonly requires a full angular scan of the specimen. Angular restrictions that exist, especially in technical processes, result in reconstruction artifacts and unknown systematic measurement errors. We investigate the use of neural networks for extrapolating the missing projection data from holographic sound pressure measurements. A bias flow liner was studied for active sound dampening in aviation. We employed a dense U-Net trained on synthetic data and compared reconstructions of simulated and measured data with and without extrapolation. In both cases, the neural network based approach decreases the mean and maximum measurement deviations by a factor of two. These findings can enable quantitative measurements in other applications suffering from limited angular access as well.
topic bias-flow liner
tomography
highspeed camera
volumetric sound pressure
dense U-Net
deep learning
url https://www.mdpi.com/2076-3417/11/10/4570
work_keys_str_mv AT oliverrothkamm denseunetforlimitedangletomographyofsoundpressurefields
AT johannesgurtler denseunetforlimitedangletomographyofsoundpressurefields
AT jurgenczarske denseunetforlimitedangletomographyofsoundpressurefields
AT robertkuschmierz denseunetforlimitedangletomographyofsoundpressurefields
_version_ 1721415342731821056

Similar Items