Image Feature Reconstruction Error for Robust Anomaly Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing anomaly detection methods struggle to accurately identify images with a high proportion of anomaly parts, often misclassifying them as normal due to high reconstruction errors.
Innovation Solution
A method involving a computer-based operation program that calculates reconstruction errors using feature quantities extracted from images, utilizing a dictionary of normal image patterns to determine anomalies by optimizing reconstruction errors through sparse coding and convolutional neural networks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If anomaly detection is performed using sparse coding with high reconstruction error thresholding, then normal images can be accurately identified, but images with high proportion of anomaly parts are misclassified as normal
Solution Approach 1:
The patent segments the anomaly detection process into multiple independent detection stages. Instead of relying on a single reconstruction error threshold, the system performs sequential detections with different thresholds and combines results, allowing it to handle both low-anomaly and high-anomaly images reliably without misclassification
Solution Approach 2:
The patent dynamically adjusts detection parameters based on image characteristics. By changing the threshold values and detection strategies according to the proportion of anomaly parts detected in each image, the system maintains high measurement precision across varying anomaly levels while improving overall detection robustness
2Productivity
If a single threshold value is used for anomaly detection, then the detection process is simple and fast, but the detection accuracy varies significantly with different anomaly proportions
Solution Approach 1:
The patent implements dynamic threshold adjustment where the detection threshold is not fixed but adapts based on the specific image being analyzed. This allows the system to maintain fast detection speeds by avoiding complex iterative processes while achieving high accuracy by selecting appropriate thresholds for each image's anomaly proportion
Solution Approach 2:
The system incorporates feedback mechanisms where detection results from initial processing stages inform subsequent detection parameters. The reconstruction error calculations and anomaly proportion assessments feed back into threshold selection, enabling accurate detection across different anomaly levels without sacrificing detection speed through multi-stage parallel processing
Data Source
AI summary
A detection method with high robustness that can determine a feature of an input image regardless of characteristics of a detected part in the image is provided. An operation program causes a computer to perform a feature quantity acquiring step of acquiring a feature quantity which is extracted from an input image, a reconstruction error calculating step of calculating a reconstruction error on the basis of a difference between the acquired feature quantity and an average which is determined in advance on the basis of a normal image, and an output step of outputting a result of calculation in the reconstruction error calculating step.