Trained Model Threshold Selection for Printed Matter Inspection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing printing technologies rely on manual threshold value selection for defect detection, leading to complex inspection results and unnecessary disposal of printed materials, as machines may incorrectly identify micro defects.
Innovation Solution
An information processing system that utilizes a trained model to automatically output a threshold value for inspection based on input print job information, including type of printed matter, image details, and apparatus information, to enhance accuracy and reduce manual intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple threshold values are prepared and manual selection is used, then inspection flexibility is improved, but operation complexity increases and automation is reduced
Solution Approach 1:
The inspection apparatus automatically determines the appropriate threshold value by comparing scan images with reference images and calculating pixel difference frequencies, eliminating the need for manual threshold selection while maintaining inspection flexibility for different defect types
Solution Approach 2:
The system dynamically adjusts the threshold value parameter based on the frequency distribution of pixel differences between scan and reference images, automatically selecting the optimal threshold for each inspection scenario rather than using fixed manual values
2Measurement precision
If low threshold value is used for inspection, then defect detection precision is improved, but false detection increases causing unnecessary disposal
Solution Approach 1:
The system uses feedback from comparing multiple scan images with reference images to determine the optimal threshold, analyzing the frequency distribution of pixel differences to set a threshold that distinguishes actual defects from normal variations
Solution Approach 2:
The system performs multiple comparisons between scan images and reference images, accumulating statistical data on pixel differences to determine the threshold, using more actions than a single comparison to achieve higher reliability
3Reliability
If high threshold value is used for inspection, then false detection is reduced, but defect detection precision decreases
Solution Approach 1:
The threshold value parameter is dynamically adjusted based on the frequency distribution of pixel differences, automatically finding the optimal balance between detecting actual defects and avoiding false detections for each inspection case
Data Source
AI summary
An information processing system includes a processor configured to, in a case where input information including print job information is input, by inputting new input information including information related to a new print job to a trained model that has been trained in advance to output information related to a threshold value for inspection performed on printed matter to be printed using the print job, output information related to a threshold value for inspection corresponding to the new input information.


