Two-field sequential color liquid crystal displays with deep learning-enabled real-time driving.

Area of Science:

• Display Technology
• Liquid Crystal Displays
• Computer Vision

Background:

• Field Sequential Color Liquid Crystal Displays (FSC-LCDs) offer advantages like tripled light efficiency and spatial resolution.
• Achieving two-field driving in FSC-LCDs is challenging due to color breakup, distortion, and computational complexity.

Purpose of the Study:

• To propose a real-time two-field FSC driving method.
• To address challenges in color breakup, distortion, and backlight crosstalk in high-resolution images.
• To provide an open-source implementation for broader accessibility.

Main Methods:

• Developed a two-field FSC driving method using two lightweight neural networks for synchronous backlight and LC signal generation.
• Implemented an over-partitioning approach to mitigate crosstalk between backlight segments.
• Utilized a GeForce RTX 3090Ti graphics card for real-time performance testing.

Main Results:

• Achieved real-time driving with network runtimes of 1.23 ms and 1.79 ms per frame.
• Demonstrated a 14.88% reduction in color breakup compared to existing methods.
• Verified low image distortion and effective crosstalk mitigation for high-resolution content.

Conclusions:

• The proposed two-field FSC driving method successfully overcomes previous limitations.
• The method offers significant improvements in color breakup and distortion while maintaining real-time performance.
• Open-source code availability facilitates further research and development in FSC-LCD technology.