Image Inspection Device Using Multi-Region Color Difference Comparison
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image inspection devices incorrectly determine the match between reference and comparison images when the orientation of the measurement target differs, assuming a fixed position and orientation.
Innovation Solution
The inspection method and device divide both reference and captured images into regions, calculate and compare color differences, and perform multiple comparisons in different combinations to accurately determine matching, even when the images are rotated.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If the position of the measurement target is assumed to be fixed during image capture, then the inspection process is simplified, but the accuracy of matching determination deteriorates when the target orientation differs from the reference image
Solution Approach 1:
The patent divides both the reference image and captured image into multiple divided regions (blocks), and performs integration processing on each region separately. This segmentation allows the system to handle orientation differences by comparing corresponding regions even when the overall image orientation varies, thus improving matching accuracy without significantly increasing overall process complexity.
Solution Approach 2:
The patent changes the parameter of image comparison from direct pixel-by-pixel comparison to comparison based on integration values (sum of pixel values) for each divided region. This parameter transformation enables the system to accommodate orientation differences while maintaining accurate matching determination.
2Measurement precision
If multiple comparison combinations are performed to handle orientation differences, then the matching determination accuracy is improved, but the inspection processing time increases
Solution Approach 1:
By dividing the image into multiple regions and calculating integration values for each region, the patent enables efficient comparison across different orientations. The segmentation allows parallel processing of multiple regions and reduces the computational burden compared to exhaustive pixel-by-pixel comparison across all possible orientations.
Solution Approach 2:
The patent performs integration processing on divided regions rather than processing the entire image as a whole. This partial action approach focuses computational resources on key regions, achieving accurate matching determination with reduced processing time compared to comprehensive full-image analysis for all orientation variations.
Data Source
AI summary
An inspection method includes: a first acquisition step of acquiring information in which information on a color difference with respect to a reference color is assigned to each of a plurality of first divided regions into which an reference image is divided; a second acquisition step of acquiring information in which information on a color difference with respect to the reference color is assigned to each of a plurality of second divided regions into which a captured image is divided; a first comparison step of comparing the information on the color difference of each of the first divided regions with the information on the color difference of each of the corresponding second divided regions; and a second comparison step of comparing the information on the color difference of each of the first divided regions with the information on the color difference of each of the second divided regions in a combination different from the combination in which the comparison is performed in the first comparison step, and whether reference image data and the captured image data match each other is determined based on a first comparison result obtained in the first comparison step and a second comparison result obtained in the second comparison step.