Image Inspection Threshold Mapping for Low-Resolution Defect Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image inspection technologies face challenges in accurately detecting defects in low-resolution images due to inappropriate setting of defect detection thresholds, leading to reduced inspection accuracy and speed, especially when dealing with regions of varying pixel value changes.
Innovation Solution
An image inspection apparatus that generates sensitivity maps with region-specific detection thresholds by averaging luminance values and dividing images into multiple regions, allowing for high-speed and high-accuracy defect detection by comparing candidate regions in inspection and non-defective images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a single defect detection threshold is used for the entire image, then the inspection process is simple and fast, but the inspection accuracy decreases in regions with varying pixel value changes
Solution Approach 1:
The patent applies local quality by dividing the inspection image into multiple regions and setting different defect detection thresholds for each region based on its specific characteristics (pixel value change magnitude). This allows each region to be inspected with an optimized threshold, improving detection accuracy without significantly increasing processing complexity.
Solution Approach 2:
The patent segments the inspection image into multiple regions and processes each region separately with region-specific thresholds. This segmentation approach enables the system to handle different image characteristics in different areas, resolving the contradiction between using a simple global threshold and achieving high local detection accuracy.
2Measurement precision
If the defect detection threshold is lowered to improve sensitivity in regions with small pixel value changes, then detection accuracy increases, but false detections increase in regions with large pixel value changes
Solution Approach 1:
The patent sets different detection thresholds for different regions based on their pixel value change characteristics. Regions with small pixel value changes use lower thresholds for high sensitivity, while regions with large pixel value changes use higher thresholds to avoid false detections, thus eliminating the harmful effects of inappropriate threshold selection.
3Measurement precision
If image resolution is increased to improve defect detection accuracy, then measurement precision improves, but processing time and computational complexity increase
Solution Approach 1:
The patent changes the parameter of defect detection threshold based on regional characteristics rather than increasing image resolution. By adjusting the threshold parameter according to pixel value change magnitude in different regions, the system achieves high detection accuracy without the computational burden of higher resolution processing.
Data Source
AI summary
An image inspection apparatus for inspecting a defect of an inspection object is disclosed. The image inspection apparatus includes an imager configured to capture an inspection image of the inspection object; and a processor configured to process the inspection image captured by the imager. The processing the inspection image by the processor includes generating a sensitivity map image which has the same size as the inspection image and a non-defective image and in which a detection threshold is set for each predetermined detection region, and detecting a defect region using the sensitivity map image.