Bad-Picture Defect Grading by Neural Network and Pattern Size
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for identifying defects in display panel production are inefficient, leading to long processing times and inaccurate judgments, which can result in missed defects and increased production costs.
Innovation Solution
A method and apparatus using a neural network model to determine defect types and sizes in images, combined with a product model to assess the impact on yield, allowing for automated grading of defects based on their type and size relative to adjacent patterns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual picture judgment is used to determine bad pictures with substantial effect on product yield, then accuracy can be maintained, but processing time becomes long and efficiency is low
Solution Approach 1:
The patent replaces the manual mechanical judgment process with an automated neural network system. The neural network model automatically analyzes bad pictures, determines defect types, calculates defect sizes, and grades defects based on predefined criteria, eliminating the need for manual intervention while maintaining high accuracy and significantly improving processing speed.
Solution Approach 2:
The system performs self-service by automatically evaluating its own performance through built-in verification mechanisms. The neural network autonomously determines defect grades without external human input, and the system includes self-verification capabilities to ensure accurate classification and grading of defects.
2Reliability
If manual picture judgment is used, then detailed analysis can be performed, but the process is time-consuming and cannot provide timely warnings
Solution Approach 1:
The patent substitutes manual analysis with an automated neural network system that processes images rapidly. The system automatically determines defect types, measures defect sizes, compares them against reference patterns, and assigns grades in real-time, providing timely warnings without sacrificing the detailed analysis capability of manual judgment.
Solution Approach 2:
The system performs preliminary actions by pre-establishing reference patterns and grading criteria before actual defect detection. The neural network uses these pre-configured references to quickly and accurately grade defects during production, eliminating the need for time-consuming manual analysis while maintaining high reliability.
3Measurement precision
If manual picture judgment is used, then comprehensive evaluation can be performed, but key process maintenance is reduced
Solution Approach 1:
The patent replaces manual evaluation with an automated neural network system that comprehensively assesses defect characteristics. The system automatically determines defect types, measures sizes, compares against reference patterns, and assigns grades, maintaining complete evaluation coverage while significantly improving productivity and enabling continuous monitoring of key processes.
Data Source
AI summary
A method and apparatus for identifying the defect grade of a bad picture, and a storage medium, which are used for solving the technical problem of it not being possible to quickly and accurately determine a bad picture that has a substantial affect on product yield. The method comprises: determining the defect size of a defect from a bad picture (101); according to a product model corresponding to the bad picture, determining the design size of a pattern corresponding to a component that is adjacent to the position of the defect (102); determining the defect grade of the defect according to the defect type of the defect and a magnitude relationship between the defect size and the design size, wherein the defect grade is the degree to which the defect affects product yield (103).