Multi-ring perfect vortex wave spatial optical communication based on convolutional neural network image recognition technology: a case study of encoded Chinese characters

This article demonstrates the first use of a multi-ring perfect vortex beam as an information carrier to encode 2,500 Chinese characters, explore the signal transmission effect under the influence of different atmospheric turbulence, and improve the convolutional neural network (CNN) algorithm for pattern recognition, which is used for information transmission in the fields of secure communication and deep space exploration. Utilize convolutional neural networks to identify damaged perfect vortex beams. Under medium turbulence intensity, the optimized ResNet-18 model can achieve 99.79% recognition accuracy, and the distortion rate of Chinese characters restored and decoded is 0.