Two-Stage Foreign Object Detection in Hyperspectral Imaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for detecting foreign objects in inspection targets using hyperspectral imaging require significant processing loads and time, especially when dealing with a large number of wavelength bands, which can be inefficient and resource-intensive.
Innovation Solution
A two-stage foreign object detection method that first identifies potential regions using a smaller band group and then confirms the presence of foreign objects using a larger band group, reducing the overall processing load and inspection time by focusing detailed analysis only on regions of interest.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If hyperspectral imaging with a large number of wavelength bands is used to detect foreign objects, then detection precision is improved, but processing load increases
Solution Approach 1:
The patent divides the detection process into two stages: first using a reduced band group to identify candidate regions, then applying full hyperspectral analysis only to those candidate regions. This segmentation approach maintains detection precision while reducing the overall processing load by avoiding unnecessary analysis of regions without foreign objects.
Solution Approach 2:
The patent applies different processing qualities to different regions of the inspection target. Regions identified as candidates receive full hyperspectral analysis (high quality processing), while other regions receive minimal or no processing. This local quality differentiation optimizes the balance between detection precision and processing efficiency.
2Measurement precision
If hyperspectral imaging with a large number of wavelength bands is used to detect foreign objects, then detection precision is improved, but inspection time increases
Solution Approach 1:
The patent segments the inspection process into a fast preliminary screening stage using reduced bands and a detailed analysis stage using full hyperspectral data only for candidate regions. This significantly reduces inspection time while maintaining detection precision by avoiding exhaustive analysis of the entire image.
Solution Approach 2:
The patent performs preliminary identification of candidate regions using a reduced band group before conducting the time-consuming full hyperspectral analysis. This preliminary action filters out regions without foreign objects, preventing wasted inspection time on areas that do not require detailed analysis.
3Reliability
If full hyperspectral analysis is applied to the entire inspection target, then detection reliability is improved, but productivity decreases
Solution Approach 1:
The patent segments the inspection target into candidate regions (identified by reduced band analysis) and non-candidate regions. Full hyperspectral analysis is applied only to candidate regions, maintaining detection reliability for foreign objects while improving productivity by processing fewer pixels with the computationally intensive full analysis.
Solution Approach 2:
The patent extracts and isolates candidate regions from the full inspection target for detailed hyperspectral analysis. This extraction approach ensures that reliable detection is applied where needed (in candidate regions) while improving overall productivity by excluding non-candidate regions from the resource-intensive analysis process.
Data Source
AI summary
A method for detecting a foreign object included in an inspection target includes acquiring first image data for the target, for which pixels each have a pixel value for a first band group including one or more wavelength bands; determining, from the first image data, one or more pixel regions that satisfy a first condition to be one or more first foreign object regions; acquiring second image data for one or more regions including the first foreign object regions, the one or more regions having pixels each of which has pixel values for a second band group including a larger number of wavelength bands than the first band group; determining, from the second image data, one or more pixel regions that satisfy a second condition to be one or more second foreign object regions, in which the foreign object is present; and outputting information regarding the second foreign object regions.