International Display Workshops General Incorporated Association

09:00 〜 09:20

[VHF4/AIS2-1] A Neural Network Based Quantitative MURA Evaluation Method Capable of Evaluating Multiple MURA on a Screen

*Satomi Kidoguchi1, Yusuke Bamba1 (1. EIZO Corporation(Japan))

Mura, Evaluation, Deep Learning, Machine Learning, Artificial Intelligence

https://doi.org/10.36463/idw.2020.0577

We developed a deep neural network-based method for evaluation of display Mura. We defined an evaluation indicator that is highly correlated with human visual evaluation. However, it could not evaluate multiple Mura on a screen properly. Therefore, we improved the evaluation indicator so that it can handle multiple Mura.