3D Light‐Field Reconstruction with Single Shot Based on Radially Self‐Accelerating Beams

Abstract Reconstructing 3D light fields from holograms mainly relies on iterative algorithms and deep learning. However, these strategies are often limited by time‐consuming and complex operations. Radially self‐accelerating beams exhibit distinct rotational characteristics during propagation, making them well‐suited for various optical systems. This paper presents an innovative approach that combines the radially self‐accelerating beams with orbital angular momentum (OAM) holography and 3D point cloud technology to enable fast and accurate 3D light‐field reconstruction from a single‐shot image. In experiments, the beams are independently convolved onto the point cloud, allowing each point to rotate around the optical axis during propagation. A light neural network is designed to deduce the relative heights of all points based on rotation properties and to reconstruct the 3D light field in 0.7 s with an accuracy of over 93%. It is anticipated that this work will provide new opportunities in the fields of 3D object measurement, real‐time particle tracking, and the innovative application of OAM holographic multiplexing.