Inpainting Versus Denoising for Dose Reduction in Scanning-Beam Microscopies

Toby Sanders, Christian Dwyer

Research output: Contribution to journalArticle

Abstract

We consider sampling strategies for reducing the radiation dose during image acquisition in scanning-beam microscopies, such as SEM, STEM, and STXM. Our basic assumption is that we may acquire subsampled image data (with some pixels missing) and then inpaint the missing data using a compressed-sensing approach. Our noise model consists of Poisson noise plus random Gaussian noise. We include the possibility of acquiring fully sampled image data, in which case the inpainting approach reduces to a denoising procedure. We use numerical simulations to compare the accuracy of reconstructed images with the 'ground truths.' The results generally indicate that, for sufficiently high radiation doses, higher sampling rates achieve greater accuracy, commensurate with the well-established literature. However, for very low radiation doses, where the Poisson noise and/or random Gaussian noise begins to dominate, then our results indicate that subsampling/inpainting can result in smaller reconstruction errors. We also present an information-theoretic analysis, which allows us to quantify the amount of information gained through the different sampling strategies and enables some broader discussion of the main results.

Original languageEnglish (US)
Article number8765610
Pages (from-to)351-359
Number of pages9
JournalIEEE Transactions on Image Processing
Volume29
DOIs
StatePublished - Jan 1 2020

Fingerprint

Dosimetry
Microscopic examination
Sampling
Scanning
Information analysis
Compressed sensing
Image acquisition
Pixels
Scanning electron microscopy
Computer simulation

Keywords

  • image denoising
  • Image sampling
  • maximum a posteriori estimation
  • scanning electron microscopy

ASJC Scopus subject areas

  • Software
  • Computer Graphics and Computer-Aided Design

Cite this

Inpainting Versus Denoising for Dose Reduction in Scanning-Beam Microscopies. / Sanders, Toby; Dwyer, Christian.

In: IEEE Transactions on Image Processing, Vol. 29, 8765610, 01.01.2020, p. 351-359.

Research output: Contribution to journalArticle

@article{1d78745ac77047a69c89ca603ee484b6,
title = "Inpainting Versus Denoising for Dose Reduction in Scanning-Beam Microscopies",
abstract = "We consider sampling strategies for reducing the radiation dose during image acquisition in scanning-beam microscopies, such as SEM, STEM, and STXM. Our basic assumption is that we may acquire subsampled image data (with some pixels missing) and then inpaint the missing data using a compressed-sensing approach. Our noise model consists of Poisson noise plus random Gaussian noise. We include the possibility of acquiring fully sampled image data, in which case the inpainting approach reduces to a denoising procedure. We use numerical simulations to compare the accuracy of reconstructed images with the 'ground truths.' The results generally indicate that, for sufficiently high radiation doses, higher sampling rates achieve greater accuracy, commensurate with the well-established literature. However, for very low radiation doses, where the Poisson noise and/or random Gaussian noise begins to dominate, then our results indicate that subsampling/inpainting can result in smaller reconstruction errors. We also present an information-theoretic analysis, which allows us to quantify the amount of information gained through the different sampling strategies and enables some broader discussion of the main results.",
keywords = "image denoising, Image sampling, maximum a posteriori estimation, scanning electron microscopy",
author = "Toby Sanders and Christian Dwyer",
year = "2020",
month = "1",
day = "1",
doi = "10.1109/TIP.2019.2928133",
language = "English (US)",
volume = "29",
pages = "351--359",
journal = "IEEE Transactions on Image Processing",
issn = "1057-7149",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

TY - JOUR

T1 - Inpainting Versus Denoising for Dose Reduction in Scanning-Beam Microscopies

AU - Sanders, Toby

AU - Dwyer, Christian

PY - 2020/1/1

Y1 - 2020/1/1

N2 - We consider sampling strategies for reducing the radiation dose during image acquisition in scanning-beam microscopies, such as SEM, STEM, and STXM. Our basic assumption is that we may acquire subsampled image data (with some pixels missing) and then inpaint the missing data using a compressed-sensing approach. Our noise model consists of Poisson noise plus random Gaussian noise. We include the possibility of acquiring fully sampled image data, in which case the inpainting approach reduces to a denoising procedure. We use numerical simulations to compare the accuracy of reconstructed images with the 'ground truths.' The results generally indicate that, for sufficiently high radiation doses, higher sampling rates achieve greater accuracy, commensurate with the well-established literature. However, for very low radiation doses, where the Poisson noise and/or random Gaussian noise begins to dominate, then our results indicate that subsampling/inpainting can result in smaller reconstruction errors. We also present an information-theoretic analysis, which allows us to quantify the amount of information gained through the different sampling strategies and enables some broader discussion of the main results.

AB - We consider sampling strategies for reducing the radiation dose during image acquisition in scanning-beam microscopies, such as SEM, STEM, and STXM. Our basic assumption is that we may acquire subsampled image data (with some pixels missing) and then inpaint the missing data using a compressed-sensing approach. Our noise model consists of Poisson noise plus random Gaussian noise. We include the possibility of acquiring fully sampled image data, in which case the inpainting approach reduces to a denoising procedure. We use numerical simulations to compare the accuracy of reconstructed images with the 'ground truths.' The results generally indicate that, for sufficiently high radiation doses, higher sampling rates achieve greater accuracy, commensurate with the well-established literature. However, for very low radiation doses, where the Poisson noise and/or random Gaussian noise begins to dominate, then our results indicate that subsampling/inpainting can result in smaller reconstruction errors. We also present an information-theoretic analysis, which allows us to quantify the amount of information gained through the different sampling strategies and enables some broader discussion of the main results.

KW - image denoising

KW - Image sampling

KW - maximum a posteriori estimation

KW - scanning electron microscopy

UR - http://www.scopus.com/inward/record.url?scp=85072755976&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85072755976&partnerID=8YFLogxK

U2 - 10.1109/TIP.2019.2928133

DO - 10.1109/TIP.2019.2928133

M3 - Article

AN - SCOPUS:85072755976

VL - 29

SP - 351

EP - 359

JO - IEEE Transactions on Image Processing

JF - IEEE Transactions on Image Processing

SN - 1057-7149

M1 - 8765610

ER -