Edge Pixel Correction for Printed Material Inspection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image inspection techniques for printed materials often fail to accurately detect defects near image edges due to misalignment and noise, leading to potential overdetection or underdetection of image defects, especially in areas with steep luminance changes.
Innovation Solution
An inspection apparatus that extracts and corrects pixel values in predetermined edge areas of both the inspection and reference images, using smoothing and noise appending processes to improve alignment and reduce false defect detection, allowing for high-accuracy detection of blots on printed materials.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the threshold value is corrected based on pixel densities in edge areas to suppress overdetection, then false defect detection is reduced, but real image defects in dark portions near edges may be overlooked
Solution Approach 1:
The image is divided into multiple areas: edge areas where luminance changes steeply and non-edge areas. Different processing is applied to each segment - threshold correction is applied selectively to edge areas to suppress false detection, while non-edge areas maintain standard detection sensitivity to ensure real defects are not missed.
Solution Approach 2:
Different threshold correction strategies are applied to different regions of the image. In edge areas with steep luminance changes, the threshold is corrected to reduce false positives. In non-edge areas, the original threshold is maintained to ensure sensitive detection of actual defects. This local differentiation resolves the contradiction between suppressing false detection and maintaining detection sensitivity.
2Measurement precision
If alignment processing is performed to reduce misalignment between reference and scanned images, then overall alignment accuracy is improved, but residual misalignment in edge areas with steep luminance changes causes large luminance differences that are detected as defects
Solution Approach 1:
The patent identifies edge areas with steep luminance changes as problematic regions where residual misalignment causes large luminance differences. By applying threshold correction specifically to these local edge areas while maintaining standard processing elsewhere, the system accounts for the harmful effect of misalignment in these specific regions without compromising overall alignment accuracy or detection sensitivity in non-edge areas.
Data Source
AI summary
From one of an inspection image of a printed material and a reference image, a predetermined area including an edge is extracted and a pixel value of a pixel is corrected, which configures the predetermined area in the inspection image or the reference image, from which the predetermined area has been extracted, and a blot on the printed material is detected by using collation results of the corrected inspection image and the reference image and a threshold value, or by using collation results of the inspection image and the corrected reference image and the threshold value.