Container Closure Inspection Using Symmetry-Guided AI Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for inspecting container closures are inefficient and lack a holistic approach to ensure targeted and precise detection of damage and positioning issues.
Innovation Solution
A computer-implemented method using a combination of machine-trained and non-machine-trained algorithms to analyze image data sets of container closures, where machine-trained algorithms identify symmetry-breaking elements and non-machine-trained algorithms verify closure positioning and damage, generating ejection signals for defective containers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a machine-trained algorithm is used to inspect all elements of the closure, then the detection accuracy is improved, but the computational time and processing complexity increase
Solution Approach 1:
The inspection process is segmented into two distinct phases: a first phase using a machine-trained algorithm to identify symmetry-breaking elements, and a second phase using the same algorithm to inspect only those identified elements for damage features. This segmentation allows the system to maintain high detection accuracy while reducing overall computational time by limiting the second inspection to only relevant elements rather than the entire closure.
Solution Approach 2:
The first machine-trained algorithm performs a preliminary action by identifying symmetry-breaking elements before the second inspection phase. This preliminary identification step filters the inspection scope, ensuring that the second algorithm only processes elements that actually require detailed inspection, thereby reducing redundant computational work while maintaining comprehensive damage detection.
2Reliability
If a comprehensive inspection of all closure elements is performed, then the inspection thoroughness is improved, but the processing complexity increases
Solution Approach 1:
The inspection methodology is divided into two sequential stages with distinct objectives. The first stage segments the problem by identifying which elements require inspection based on symmetry breaking. The second stage segments the actual damage inspection to only those identified elements. This dual segmentation reduces processing complexity while maintaining inspection thoroughness through a structured two-phase approach.
Solution Approach 2:
The first algorithm performs a preliminary classification action that simplifies the subsequent inspection process. By pre-identifying symmetry-breaking elements, the system reduces the complexity of the second inspection phase, transforming a potentially complex full-element inspection into a focused examination of only relevant elements, thereby maintaining thoroughness with reduced complexity.
3Productivity
If only symmetry-breaking elements are inspected, then the processing efficiency is improved, but the detection coverage may be reduced
Solution Approach 1:
The invention leverages the asymmetry principle by focusing inspection on symmetry-breaking elements, which are inherently asymmetric features on the closure. These elements are precisely the locations where damage is most likely to occur and where sealing integrity is most critical. By targeting asymmetric features rather than uniformly inspecting all elements, the system achieves both high processing efficiency and comprehensive detection coverage.
Solution Approach 2:
The inspection system applies local quality by concentrating detailed damage inspection resources on specific high-risk locations (symmetry-breaking elements) rather than uniformly distributing inspection effort across all closure elements. This localized approach ensures that areas most critical to sealing integrity receive the most thorough inspection, maintaining detection coverage while improving overall processing efficiency.
Data Source
Figure 1
Figure 2~3
AI summary
A computer-implemented method (100) for inspecting at least one closure of a container extending along a longitudinal axis through a container mouth, comprising the following steps: receiving (102) at least one spatially resolved image data set of a closed container; analyzing (104) the image data set with a first machine-trained algorithm to determine whether the closure has at least one element that breaks a rotational symmetry of the closure about the longitudinal axis; if the element is present (106): inspecting (108) at least the element using a second machine-trained algorithm to determine whether a damage feature is present; if a damage feature is present (110): providing (112) an ejection signal for the container;If no element is present (114): inspecting (116) the closure with a first non-machine-trained algorithm to determine whether the closure is correctly positioned on the container and/or is damaged; if the closure is damaged and/or incorrectly positioned on the container (118): providing (112) an ejection signal for the container. The method (100) can perform the inspection of the container closures with increased efficiency, so that the inspection of the closure can be targeted and holistic.