Diffusion probabilistic models for compressive SAR imaging

Odysseas Pappas; Perla Mayo; Andrew Austin; Alin Achim

doi:10.1117/12.3053636

Diffusion probabilistic models for compressive SAR imaging

Research output: Chapter in Book/Report/Conference proceeding › Conference Contribution (Conference Proceeding)

32 Downloads (Pure)

Abstract

Compressed sensing Synthetic Aperture Radar (SAR) image formation, formulated as an inverse problem and solved with traditional iterative optimization methods can be very computationally expensive. We investigate the use of denoising diffusion probabilistic models for compressive SAR image reconstruction, where the diffusion model is guided by a poor initial reconstruction from sub-sampled data obtained via standard imaging methods. We present results on real SAR data and compare our compressively sampled diffusion model reconstruction with standard image reconstruction methods utilizing the full data set, demonstrating the potential performance gains in imaging quality.

Original language	English
Title of host publication	Proceedings of SPIE
Subtitle of host publication	Algorithms for Synthetic Aperture Radar Imagery XXXII
Editors	Edmund Zelnio, Frederick D. Garber
Publisher	SPIE - The International Society for Optical Engineering
Chapter	1345607
Number of pages	6
ISBN (Electronic)	9781510687028
ISBN (Print)	9781510687011
DOIs	https://doi.org/10.1117/12.3053636
Publication status	Published - 28 May 2025
Event	SPIE Defense + Commercial Sensing 2025 - Orlando, United States Duration: 13 Apr 2025 → 17 Apr 2025

Publication series

Name	Proceedings of SPIE
Publisher	SPIE
Volume	13456
ISSN (Print)	0277-786X
ISSN (Electronic)	1996-756X

Conference

Conference	SPIE Defense + Commercial Sensing 2025
Country/Territory	United States
City	Orlando
Period	13/04/25 → 17/04/25

Bibliographical note

Publisher Copyright:
© COPYRIGHT SPIE.

Keywords

Compressed Sensing
Diffusion Denoising Probabilistic Models
SAR Imaging
Machine Learning

Access to Document

10.1117/12.3053636

Full-text PDF (accepted manuscript)
This is the accepted author manuscript (AAM) of the article which has been made Open Access under the University of Bristol's Scholarly Works Policy. The final published version (Version of Record) can be found on the publisher's website. The copyright of any third-party content, such as images, remains with the copyright holder.
Accepted author manuscript, 1.35 MBLicence: CC BY

Handle.net

Persistent link

Cite this

@inproceedings{9f17816f0b4a4adcb4d3e121bc5cbb21,

title = "Diffusion probabilistic models for compressive SAR imaging",

abstract = "Compressed sensing Synthetic Aperture Radar (SAR) image formation, formulated as an inverse problem and solved with traditional iterative optimization methods can be very computationally expensive. We investigate the use of denoising diffusion probabilistic models for compressive SAR image reconstruction, where the diffusion model is guided by a poor initial reconstruction from sub-sampled data obtained via standard imaging methods. We present results on real SAR data and compare our compressively sampled diffusion model reconstruction with standard image reconstruction methods utilizing the full data set, demonstrating the potential performance gains in imaging quality. ",

keywords = "Compressed Sensing, Diffusion Denoising Probabilistic Models, SAR Imaging, Machine Learning",

author = "Odysseas Pappas and Perla Mayo and Andrew Austin and Alin Achim",

note = "Publisher Copyright: {\textcopyright} COPYRIGHT SPIE.; SPIE Defense + Commercial Sensing 2025 ; Conference date: 13-04-2025 Through 17-04-2025",

year = "2025",

month = may,

day = "28",

doi = "10.1117/12.3053636",

language = "English",

isbn = "9781510687011",

series = "Proceedings of SPIE",

publisher = "SPIE - The International Society for Optical Engineering",

editor = "Edmund Zelnio and Garber, \{Frederick D.\}",

booktitle = "Proceedings of SPIE",

}

Pappas, O, Mayo, P, Austin, A & Achim, A 2025, Diffusion probabilistic models for compressive SAR imaging. in E Zelnio & FD Garber (eds), Proceedings of SPIE: Algorithms for Synthetic Aperture Radar Imagery XXXII., 1345607, Proceedings of SPIE, vol. 13456, SPIE - The International Society for Optical Engineering, SPIE Defense + Commercial Sensing 2025, Orlando, Florida, United States, 13/04/25. https://doi.org/10.1117/12.3053636

Diffusion probabilistic models for compressive SAR imaging. / Pappas, Odysseas; Mayo, Perla; Austin, Andrew et al.
Proceedings of SPIE: Algorithms for Synthetic Aperture Radar Imagery XXXII. ed. / Edmund Zelnio; Frederick D. Garber. SPIE - The International Society for Optical Engineering, 2025. 1345607 (Proceedings of SPIE; Vol. 13456).

Research output: Chapter in Book/Report/Conference proceeding › Conference Contribution (Conference Proceeding)

TY - GEN

T1 - Diffusion probabilistic models for compressive SAR imaging

AU - Pappas, Odysseas

AU - Mayo, Perla

AU - Austin, Andrew

AU - Achim, Alin

PY - 2025/5/28

Y1 - 2025/5/28

N2 - Compressed sensing Synthetic Aperture Radar (SAR) image formation, formulated as an inverse problem and solved with traditional iterative optimization methods can be very computationally expensive. We investigate the use of denoising diffusion probabilistic models for compressive SAR image reconstruction, where the diffusion model is guided by a poor initial reconstruction from sub-sampled data obtained via standard imaging methods. We present results on real SAR data and compare our compressively sampled diffusion model reconstruction with standard image reconstruction methods utilizing the full data set, demonstrating the potential performance gains in imaging quality.

AB - Compressed sensing Synthetic Aperture Radar (SAR) image formation, formulated as an inverse problem and solved with traditional iterative optimization methods can be very computationally expensive. We investigate the use of denoising diffusion probabilistic models for compressive SAR image reconstruction, where the diffusion model is guided by a poor initial reconstruction from sub-sampled data obtained via standard imaging methods. We present results on real SAR data and compare our compressively sampled diffusion model reconstruction with standard image reconstruction methods utilizing the full data set, demonstrating the potential performance gains in imaging quality.

KW - Compressed Sensing

KW - Diffusion Denoising Probabilistic Models

KW - SAR Imaging

KW - Machine Learning

U2 - 10.1117/12.3053636

DO - 10.1117/12.3053636

M3 - Conference Contribution (Conference Proceeding)

SN - 9781510687011

T3 - Proceedings of SPIE

BT - Proceedings of SPIE

A2 - Zelnio, Edmund

A2 - Garber, Frederick D.

PB - SPIE - The International Society for Optical Engineering

T2 - SPIE Defense + Commercial Sensing 2025

Y2 - 13 April 2025 through 17 April 2025

ER -

Diffusion probabilistic models for compressive SAR imaging

Abstract

Publication series

Conference

Bibliographical note

Keywords

Access to Document

Handle.net

Fingerprint

Cite this