Logo image
Res-U2Net: untrained deep learning for phase retrieval and image reconstruction
Journal article   Peer reviewed

Res-U2Net: untrained deep learning for phase retrieval and image reconstruction

Carlos Osorio Quero, Daniel Leykam and Irving Rondon Ojeda
Journal of the Optical Society of America. A, Optics, image science, and vision, Vol.41(5), pp.766-773
01/05/2024
PMID: 38856563

Abstract

Optics Physical Sciences Science & Technology
Conventional deep learning -based image reconstruction methods require a large amount of training data, which can be hard to obtain in practice. Untrained deep learning methods overcome this limitation by training a network to invert a physical model of the image formation process. Here we present a novel, to our knowledge, untrained Res-U2Net model for phase retrieval. We use the extracted phase information to determine changes in an object's surface and generate a mesh representation of its 3D structure. We compare the performance of Res-U2Net phase retrieval against UNet and U2Net using images from the GDXRAYdataset. (c) 2024 Optica Publishing Group

Metrics

1 Record Views

Details

Logo image