Simultaneous recognition of the orbital angular momentum and phase singularity position of multiple vortex beams from speckle pattern via deep learning

Abstract

The orbital angular momentum (OAM) carried by vortex beams can be used for encoding information in optical communications. The measurement of OAM is critical for these applications. However, traditional methods of measuring OAM are invalid in the presence of a scattering medium because the spiral wavefront of a vortex beam is distorted. To address this issue, we propose a deep learning-based method to simultaneously recognize the OAM and the position of the phase singularity from the speckle pattern generated by the propagation of vortex beams through a scattering medium. Our method achieves an accuracy of more than 98 % in identifying both the topological charge and singularity position information of multi-beam vortex beams with changing positions using deep learning technology. We also explore the minimum displacement accuracy that can be recognized by the neural network. When the displacement of the vortex beam with changing position is reduced to 1 pixel (12.5 µm), the information recognition accuracy remains above 95 %. Our method is a promising approach for accurate and efficient measurement of OAM and phase singularity position in the presence of scattering media.