P‐101: Convolutional Neural Network to Optimize Dual‐Panel Display

Abstract

Liquid crystal displays (LCDs) have the disadvantage of light leakage, resulting in low contrast ratio (CR). Dual‐layer LCDs can greatly improve CR by bonding two layers of LC panels. However, there is a certain distance between the two LC panels, and the ghost may appear when viewing off‐axis. At present, the solutions for this problem cannot achieve a good balance on processing time and display quality. In this paper, convolutional neural network (CNN) is used to optimize dual‐layer display as well as control processing time. The residual blocks are included in the network to improve the quality of the output image. The two output images that correspond to two layers of LC panels are reconstructed from different viewing angles to make comparisons with other methods in terms of processing time and display quality. The simulation results show that compared with other methods, the proposed CNN method can greatly reduce the processing time while maintaining high display quality, which presents high practicability.