Reconstruction of fractional vortex phase evolution by generative adversarial networks.

Abstract

Digital signal coding based on the combination of vortex beam orbital angular momentum (OAM) and vortex optical phase information has made many achievements in optical communication. The accuracy of the vortex optical phase is the key to improving the efficiency of communication coding. In this regard, we propose a depth learning model based on the generative adversarial network (GAN) to accurately recover the phase image information of fractional vortex patterns at any diffraction distance, thus solving the problem that it is difficult to determine the phase information of fractional vortex patterns at different transmission distances due to the phase evolution. Compared with other depth learning methods, the phase recovery result of GAN is not affected by the diffraction distance, which is the first time we know that this method is applied to the fractional order optical vortex. Our work provides a new idea for the accurate identification of multi-singular structured light.