Mura Defect Detection via Edge Feature Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current display defect detection methods, particularly for Mura defects in display panels, face challenges such as high error rates and reduced accuracy due to large numbers of arithmetic operations and subjective worker determinations, especially when detecting repetitive patterns.
Innovation Solution
A display defect detection system and method that preprocesses images by resizing based on Mura size, detecting edge components, removing non-repetitive feature values, and calculating final feature values to accurately locate Mura defects, reducing arithmetic operations and enhancing detection accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional Mura detection algorithms perform block-unit and sub-block-unit arithmetic operations based on contrast ratio, standard deviation, and amplitude, then comprehensive feature analysis is achieved, but the number of arithmetic operations increases and detection accuracy decreases for repetitive Mura patterns
Solution Approach 1:
The patent divides the panel image into multiple blocks and further divides each block into sub-blocks, then performs arithmetic operations on each sub-block separately. This segmentation allows the system to handle repetitive Mura patterns more effectively by analyzing local variations within each sub-block rather than treating the entire image as a single unit, thereby improving detection accuracy while managing computational complexity through structured division.
Solution Approach 2:
The patent calculates multiple parameters including contrast ratio, standard deviation, and amplitude for each sub-block, and dynamically adjusts the detection threshold based on the statistical distribution of these parameters across all sub-blocks. By changing the parameters from fixed thresholds to dynamically adjusted values based on local statistics, the system achieves higher accuracy for repetitive patterns without requiring an excessive number of arithmetic operations.
2Reliability
If worker directly detects Mura defects with eyes, then subjective determination is performed, but detection rate of fine defects decreases and eyes become easily fatigued
Solution Approach 1:
The patent replaces the human visual inspection system with an automated computational system that performs arithmetic operations on digital image data. Instead of relying on human eyes and subjective judgment, the system uses objective calculations of contrast ratio, standard deviation, and amplitude to detect Mura defects, thereby eliminating eye fatigue and subjective variability while improving fine defect detection capability through precise numerical analysis.
3Measurement precision
If block-unit and sub-block-unit arithmetic operations are performed to detect Mura, then comprehensive analysis is achieved, but error of Mura detection increases when Mura occurs in two or more adjacent blocks
Solution Approach 1:
The patent merges the detection results from multiple adjacent sub-blocks by comparing their calculated parameters (contrast ratio, standard deviation, amplitude) and identifying patterns that span across block boundaries. When Mura occurs in two or more adjacent blocks, the system combines the arithmetic operation results from these blocks to detect the extended defect pattern, thereby reducing detection errors for multi-block Mura occurrences while maintaining comprehensive analysis capability.
Data Source
AI summary
A display defect detection circuit of a display defect detection system includes a preprocessing circuit configured to receive a capture image of a test pattern, displayed by a display panel, as a panel image including target Mura having a repetitive characteristic and preprocess the panel image to output a preprocessing image and a Mura detection circuit configured to decrease a total size of the preprocessing image on the basis of a size of the target Mura to generate a resize image, detect an edge component in the resize image to generate an edge map image having feature values, remove a feature value of a non-repetitive type among the feature values of the edge map image to generate a feature map image, and detect a display position of the target Mura on the basis of a final feature value of the target Mura calculated based on the feature map image.