Implicit Neural Representations for Deconvolving SAS Images

Albert Reed, Thomas Blanford, Daniel C. Brown, Suren Jayasuriya

Research output: Chapter in Book/Report/Conference proceedingConference contribution

4 Scopus citations


Synthetic aperture sonar (SAS) image resolution is constrained by waveform bandwidth and array geometry. Specifically, the waveform bandwidth determines a point spread function (PSF) that blurs the locations of point scatterers in the scene. In theory, deconvolving the reconstructed SAS image with the scene PSF restores the original distribution of scatterers and yields sharper reconstructions. However, deconvolution is an ill-posed operation that is highly sensitive to noise. In this work, we leverage implicit neural representations (INRs), shown to be strong priors for the natural image space, to deconvolve SAS images. Importantly, our method does not require training data, as we perform our deconvolution through an analysis-by-synthesis optimization in a self-supervised fashion. We validate our method on simulated SAS data created with a point scattering model and real data captured with an in-air circular SAS. This work is an important first step towards applying neural networks for SAS image deconvolution.

Original languageEnglish (US)
Title of host publicationOCEANS 2021
Subtitle of host publicationSan Diego - Porto
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9780692935590
StatePublished - 2021
EventOCEANS 2021: San Diego - Porto - San Diego, United States
Duration: Sep 20 2021Sep 23 2021

Publication series

NameOceans Conference Record (IEEE)
ISSN (Print)0197-7385


ConferenceOCEANS 2021: San Diego - Porto
Country/TerritoryUnited States
CitySan Diego

All Science Journal Classification (ASJC) codes

  • Oceanography


Dive into the research topics of 'Implicit Neural Representations for Deconvolving SAS Images'. Together they form a unique fingerprint.

Cite this