A deep learning-based electronic screen defect detection method

By combining three-color channel information and pixel spatial location with deep learning methods, the problem of misjudgment in traditional electronic screen detection methods under complex display conditions is solved, and stable and reliable defect category identification and recording are achieved.

CN122415541APending Publication Date: 2026-07-17JIANGXI BLUEPRINT INTELLIGENT MANUFACTURING TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
JIANGXI BLUEPRINT INTELLIGENT MANUFACTURING TECHNOLOGY CO LTD
Filing Date
2026-04-24
Publication Date
2026-07-17

AI Technical Summary

Technical Problem

Traditional electronic screen defect detection methods struggle to reliably identify defect categories under complex display conditions, resulting in unreliable detection results, especially prone to misjudgment in areas of brightness transition or background color gradient.

Method used

A deep learning-based approach is adopted, which combines convolutional neural networks and deep learning defect recognition networks with three-color channel information and pixel spatial location to extract brightness distribution features and color change features, identify and classify defect areas, and form a defect category distribution map.

Benefits of technology

It achieves stable defect category identification under complex display conditions, provides complete location marking and category records, and improves the reliability and consistency of detection results.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122415541A_ABST
    Figure CN122415541A_ABST
Patent Text Reader

Abstract

The present application relates to the technical field of electronic screen defect detection, in particular to an electronic screen defect detection method based on deep learning, which acquires sub-screen gray scale lighting pictures, collects three-color channel information to construct luminance distribution image data, inputs a convolutional neural network to extract luminance and color features, identify abnormal pixels and form defect area features, inputs a deep learning network to construct defect sample expression and identify defect categories, establishes defect spatial distribution and generates electronic screen defect labeling results. The present application controls sub-screen pixel array gray scale lighting through a drive board card and collects red, green and blue channel information, so that pixel luminance and spatial position form a unified distribution structure, feature extraction is performed in combination with color change and pixel distribution relationship, abnormal pixels form area expression according to spatial adjacency relationship, luminance change and color change are analyzed cooperatively, subtle luminance fluctuation and color shift areas are distinguished, and defect categories and pixel positions form corresponding labeling structures.
Need to check novelty before this filing date? Find Prior Art