Suppressing spatio-temporal artifacts in field-sequential color LCD-TVs via a dual-mixed-blue architecture and lightweight deep learning

Field-sequential color (FSC) LCDs triple optical efficiency and spatial resolution but suffer from spatio-temporal artifacts: color breakup (CBU), distortion, and flicker. While current FSC algorithms, such as the Stencil approach, effectively balance CBU and distortion, they inherently conflict with flicker suppression, which is a critical limitation for large-format FSC-LCD TVs where flicker is exacerbated. Furthermore, the lack of standardized methods for chromatic flicker assessment continues to hinder FSC-LCD development. To address these challenges, this study first introduces a Color-Resolved Flicker Visibility Metric (CR-FVM) based on the DKL opponent-color space to quantify flicker in FSC-LCDs. Next, a Dual-Mixed-Blue field-sequential architecture (DMB-FSC) is proposed. By utilizing two mixed-color subframes and a residual blue subframe, DMB-FSC doubles the parameter space for multi-objective optimization (MOO) compared to the Stencil approach, enabling the MOO algorithm to achieve a 3-objective Pareto optimality. Finally, a hardware-friendly Nano RepVGG network is developed to infer optimal backlight signals, bypassing the MOO iteration, achieving real-time backlight prediction in just 0.65 ms. Simulations and experiments on an 85-inch FSC-LCD TV demonstrate state-of-the-art CBU suppression and color fidelity while substantially reducing flicker perception.