Res-U2Net: Untrained Deep Learning for Phase Retrieval and Image Reconstruction
This addresses the challenge of data scarcity in deep learning-based image reconstruction for applications like 3D surface analysis, though it is incremental as it builds on existing untrained methods.
The paper tackled the problem of phase retrieval for 3D image reconstruction without requiring large training datasets by introducing an untrained Res-U2Net model, achieving performance improvements over UNet and U2Net on the GDXRAY dataset.
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 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 GDXRAY dataset.