Machine learning phase modulation of liquid crystal devices for three-dimensional display

Abstract

A machine learning phase modulation scheme based on convolutional neural networks (CNN) and recurrent neural network (RNN) is proposed to carry out the regression task of liquid crystal (LC) device electric field prediction for the 2D/3D switchable display. The hybrid neural network is built and trained based on the illuminance distribution under three-dimensional (3D) display. Compared with manual phase modulation, the modulation method using a hybrid neural network can achieve higher optical efficiency and lower crosstalk in the 3D display. The validity of the proposed method is confirmed through simulations and optical experiments.