58‐1: A Novel On‐line, Fast Color Correction by Machine Learnings

Abstract

A new efficient color correction method, utilizing neural network models, is proposed by this study for OLED panels in production, offering a transformative approach to achieve accurate and rapid color calibration. Utilizing transfer learning with a limited dataset, it creates color conversion matrices for AI‐driven color correction. The process involves selecting a similar baseline model from prior batches with a ΔE00 below 1, then refining it using a representative dataset. This model corrects white point and overall color, achieving a ΔE00 less than 2. It corrects 300 data points in less than 4 seconds, demonstrating both speed and accuracy.