Metal Strip Image Correction for Abnormality Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing technologies for metal strips struggle with accuracy in abnormality detection due to variations in illuminating conditions and camera capture conditions, which affect the luminance values of images.
Innovation Solution
A method and apparatus that correct image luminance by selecting a reference image with optimal luminance values, calculating luminance differences, and adjusting the luminance of subsequent images to match the reference, thereby stabilizing threshold values for abnormality detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If image processing is performed under varying illuminating conditions and camera capture conditions, then the system can operate in different environments, but the luminance values vary causing decreased accuracy in abnormality detection
Solution Approach 1:
The patent applies parameter changes by dynamically adjusting the threshold value based on the actual luminance characteristics of each image. Instead of using a fixed threshold, the system calculates the threshold as a proportion of the standard deviation of luminance values in the image, allowing the detection parameters to adapt to varying illuminating conditions and camera settings while maintaining detection accuracy
Solution Approach 2:
The patent implements preliminary action by calculating statistical parameters (mean and standard deviation of luminance values) from the image data before performing abnormality detection. This preprocessing step establishes a baseline for the current imaging conditions, enabling the subsequent threshold to be set appropriately for each specific image rather than using a predetermined fixed value
2Productivity
If fixed threshold values are used for abnormality detection, then the processing is simple and fast, but the accuracy decreases when illuminating conditions vary
Solution Approach 1:
The patent implements dynamics by making the threshold value dynamic rather than fixed. The threshold is calculated in real-time for each image based on the standard deviation of luminance values, allowing the system to automatically adapt to changing lighting conditions and camera settings. This dynamic approach maintains high processing speed while significantly improving detection accuracy across varying conditions
Solution Approach 2:
The patent applies feedback by using the statistical characteristics (standard deviation) of the actual image data to adjust the threshold value. The system measures the luminance variation in the image and uses this feedback information to set an appropriate threshold, creating a closed-loop system that automatically adapts to different imaging conditions without requiring manual recalibration
Data Source
Figure 1
Figure 2(A)~2(B)
Figure 3
AI summary
The present invention provides the accuracy in abnormality detection by image processing of a strip even where the illumination condition or the imaging condition of a camera is different, an acquisition section 81a acquiring a plurality of images of a surface of a metal strip 1 in a steady state, which are captured in different illumination conditions or different imaging conditions, a selection section 81b selecting a reference pixel point/range included in a portion indicative of the metal strip 1 from within the plurality of images, a determination section 81c determining, from among the plurality of images, a first image in which the luminance of any one of an R value, a G value, and a B value in average in the reference pixel point/range is highest, a calculation section 81d calculating luminance differences, for all of images other than the first image, by subtracting an R value, a G value, and a B value in average in the reference pixel point/range of another image other than the first image from the R value, the G value, and the B value in average in the reference pixel point/range of the first image, respectively, and a correction section 81e adding or subtracting an absolute value of each luminance difference to or from corresponding one of a R value, a G value, and a B value at all of the pixel points in all of the images other than the first image.