Method for detecting surface defects of organic light emitting diode

By combining high and low frequency generation models, generating maps and uncertainty maps are output by low-frequency and high-frequency generators, and the global uncertainty map is fused and the defect features are enhanced. This solves the problems of difficulty in capturing multi-scale defect features and environmental interference in the detection of organic light-emitting diode display panels, and improves the detection accuracy and efficiency.

CN122391815APending Publication Date: 2026-07-14GUANGDONG UNIV OF TECH
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
GUANGDONG UNIV OF TECH
Filing Date
2026-03-04
Publication Date
2026-07-14

AI Technical Summary

Technical Problem

Existing surface defect detection technologies for organic light-emitting diode (OLED) display panels struggle to simultaneously and accurately capture multi-scale defect features, are susceptible to imaging environment interference leading to false detections, and suffer from low detection accuracy and efficiency.

Method used

A high-low frequency combined generation model is adopted, including a low-frequency generator and a high-frequency generator. The low-frequency generator and the high-frequency generator output the generated map and uncertainty map of the corresponding frequency band respectively. The map is fused to obtain a global uncertainty map. The defect features are enhanced by combining the defect-free image and the original image. Finally, the map is input into the improved defect detection network.

Benefits of technology

It effectively improves the accuracy and efficiency of surface defect identification and detection, and solves the problems of missed detection and false detection caused by the interference of panel texture and the lack of obvious defect features in traditional detection methods.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122391815A_ABST
    Figure CN122391815A_ABST
Patent Text Reader

Abstract

The application provides an organic light-emitting diode surface defect detection method, first, a trained low-frequency and high-frequency generator is used to respectively output a generated graph and an uncertainty graph corresponding to a frequency band, a global uncertainty graph is obtained through fusion, an initial residual error graph is obtained by combining a defect-free image and an original organic light-emitting diode panel image, and defect features are strengthened, finally, the strengthened residual error graph, the original image and the global uncertainty graph are input into an improved defect detection network, the recognition precision and the detection efficiency of the surface defects are effectively improved, and the problems that a traditional detection method is prone to panel texture interference and defect features are not obvious, leading to missed detection and misdiagnosis are solved.
Need to check novelty before this filing date? Find Prior Art