International Display Workshops General Incorporated Association

13:50 〜 14:10

[DES6-2 (Invited)] A Cost-effective Burn-in Compensation Method Using Deep Convolutional Networks with Detail Layer Accumulation

*Jiheon Ok1、Un-Ki Park1、Sewhan Na1、Hyeon-Su Park1、Hyun-Wook Lim1、Jae-Youl Lee1 (1.Samsung Electronics (Korea))

OLED burn-in, Data-counting, Deep convolutional network, Multi-unit compensation

https://doi.org/10.36463/idw.2022.1012

In this paper, a cost-effective burn-in compensation method based on optical imaging is proposed. A deep convolutional networks using the accumulated detail layer as a reduced reference is applied to the captured images. Experimental results show that the proposed method reconstructs burn-in details effectively to be followed by multi-unit compensation.