Product Image Anomaly Detection with Extreme Value Thresholds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting anomalies in digital images of products require a large number of digital images for a learning process to set a threshold value, leading to inefficiencies in production time and uncertainty in determining the required number of images.
Innovation Solution
A method that uses statistical distributions, specifically extreme value distributions, to determine threshold values for anomaly detection based on a smaller number of digital images, allowing for quick and reliable threshold setting by parameterizing probability density functions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a large number of digital images are used for the learning process to set threshold values, then the reliability of anomaly detection is improved, but the production time is increased and productivity is reduced
Solution Approach 1:
The patent applies preliminary action by performing the learning process and threshold determination automatically during product changeovers or setup phases, before actual production begins. This allows the system to be pre-trained with sufficient images without impacting ongoing production productivity, as the threshold values are established in advance for subsequent anomaly detection.
Solution Approach 2:
The patent replaces manual threshold setting and manual image selection with an automated computer-implemented learning process. The system automatically selects images, processes them, and determines optimal threshold values using statistical methods, eliminating the need for manual intervention and reducing the time required while maintaining or improving reliability.
2Measurement precision
If a large number of digital images are collected for the learning process, then the accuracy of threshold determination is improved, but the complexity of the process increases
Solution Approach 1:
The patent changes the parameter of image selection from random or manual selection to automated selection based on specific criteria (e.g., image quality, representativeness). The system automatically filters and selects only the most suitable images for the learning process, reducing the total number of images needed while maintaining or improving threshold determination accuracy.
Solution Approach 2:
The patent uses digital copies and processing of image data rather than physical handling of numerous images. The computer-implemented system processes digital representations, allowing efficient storage, retrieval, and analysis of large numbers of images without increasing physical complexity, and enables automated statistical analysis that improves accuracy.
3Ease of operation
If manual threshold setting is used, then the process is simple to understand, but the reliability of anomaly detection decreases
Solution Approach 1:
The patent implements self-service by enabling the system to automatically determine optimal threshold values through a computer-implemented learning process. The system autonomously analyzes images, applies statistical methods, and sets thresholds without requiring manual intervention, thereby maintaining operational simplicity while significantly improving detection reliability through data-driven automation.
Solution Approach 2:
The patent incorporates feedback mechanisms where the system continuously learns from analyzed images and adjusts threshold values based on statistical analysis results. This automated feedback loop allows the system to improve its anomaly detection capability over time while keeping the user interface simple, as operators only need to initiate the learning process rather than manually tune complex parameters.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables efficient and accurate anomaly detection with reduced image requirements, minimizing production downtime and improving the reliability of anomaly identification.
Implementation Method 1
inspection devices are used in practice which irradiate the product with electromagnetic radiation, in particular with radiation in the X-ray spectrum
Implementation Method 2
anomalies to be detected can lead to image areas during transmission that have a higher or lower 'grey value'
Implementation Method 3
this 'reflection' is physically caused by a scattering of the radiation penetrating into a volume region
Implementation Method 4
or by generating a fluorescent radiation in the volume region
Data Source
AI summary
A method for detecting anomalies in digital images of products, wherein a region of a digital image is detected as a maximum anomaly if the value of a property of the region is greater than a predetermined maximum threshold, and/or wherein a region is detected as a minimum anomaly if the value of the property of the region is less than a predetermined minimum threshold. The maximum threshold value and/or the minimum threshold value are determined in a learning process using relatively few digital images based on a statistical distribution of the largest or smallest values of a specific quantity used for the detection of anomalies in digital images to be examined.