Untrained network regularized by total variation in single-shot lensless holography

• DOI: 10.1016/j.rinp.2023.107174
• Publication Date: 2023-11-10T06:03:17Z
• Authors: Yifan Feng; Jie Xu; Jiannan Jiao; Liyun Zhong; Xiaoxu Lu; Jindong Tian
• Keywords: Net (polyhedron); Single shot

The optical complex-amplitude (CA) distribution of an object contains rich information, providing insights into the object's characteristics such as retardation and absorption. Coaxial lensless holography (CLH) using a learning-based approach offers promise for retrieving CA maps with advantages compact setup single-shot acquisition, while suffers from laborious time-consuming acquisition datasets labels required network training. To address this challenge, we propose untrained neural Lp-norm total variation regularization (LTVR-net) by integrating physical model learning process. LTVR-net effectively suppresses twin-image artifact noises in reconstructing images, outperforming traditional methods on quantitative metrics. Besides, retrieval results at different imaging distances consistently exhibit excellent performance, indicating that possesses distance-resolution-balanced characteristic. This feature holds expanding application scope CLH, allowing more versatile flexible configurations various scenarios. Furthermore, experimental biological tissue demonstrate ability to reveal fine structures clear boundaries, highlighting its superiority imaging. These collectively prove is untrained, single-shot, distance-robust capable achieving high-quality retrieval.