Detection Method for OLED Pixel Defects Based on Extended Feature Pyramid

Fig. 4. Typical 6 types of samples. (a) Without defect; (b) single class satellite points; (c) multi class satellite points; (d) incomplete bank; (e) large droplet of ink overflows bank; (f) multiple defects

Download full size

Fig. 5. Augmentation examples of inkjet printing OLED pixel image data. (a) Original image; (b) random sharpness; (c) flip transformation; (d) random contrast; (e) random cropping; (f) random brightness

Download full size

Fig. 6. Effect of different methods on defect segmentation in inkjet printing OLED pixel image data set. (a) Original images; (b) ground truths; (c) predicted heat maps; (d) predicted binary masks; (e) defect segmentation results

Download full size

Fig. 7. Segmentation effect of proposed method on each defect. (a) Original images; (b) ground truths; (c) predicted heat maps;

Download full size

Table 1. Sample distribution of OLED pixel data set for inkjet printing
View table
Table 1. Sample distribution of OLED pixel data set for inkjet printing
Data set
Satellite
point
Incomplete bank Overflow Multiple defects Normal Total
Train 0 0 0 0 860 860
Test 45 38 48 53 40 224
Total 45 38 48 53 900 1084

Table 2. Performance index of defect detection for each category of inkjet printed OLED pixel image

View table

Table 2. Performance index of defect detection for each category of inkjet printed OLED pixel image

Defect type	Image level AUC /%	Pixel level AUC /%
Average	99.8	88.8
Single class satellite points	100.0	96.5
Multi class satellite points	100.0	85.0
Incomplete bank	99.8	95.1
Large droplet of ink overflows bank	99.9	83.9
Multiple defects	99.3	83.5

Table 3. Comparison of defect detection effects of different methods on inkjet printed OLED data sets
View table
Table 3. Comparison of defect detection effects of different methods on inkjet printed OLED data sets
Algorithm Image level AUC /% Pixel level AUC /%
SPADE 85.5 79.7
PaDiM 95.3 84.2
ResNet18+FPN 94.3 85.1
ResNet18+Expanded FPN 99.8 88.8

Tools

Get Citation

Copy Citation Text

Lan Liu, Yun Ye, Tailiang Guo. Detection Method for OLED Pixel Defects Based on Extended Feature Pyramid[J]. Acta Optica Sinica, 2023, 43(2): 0212005

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: Instrumentation, Measurement and Metrology

Received: Jul. 4, 2022

Accepted: Jul. 25, 2022

Published Online: Feb. 7, 2023

The Author Email: Ye Yun (yeyun07@fzu.edu.cn)

DOI:10.3788/AOS221411

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology

Table 1. Sample distribution of OLED pixel data set for inkjet printing

Table 1. Sample distribution of OLED pixel data set for inkjet printing

Table 2. Performance index of defect detection for each category of inkjet printed OLED pixel image

Table 2. Performance index of defect detection for each category of inkjet printed OLED pixel image

Table 3. Comparison of defect detection effects of different methods on inkjet printed OLED data sets

Table 3. Comparison of defect detection effects of different methods on inkjet printed OLED data sets