Document Image Anomaly Detection Using ROI-Specific Thresholds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing digital identity verification systems face challenges in detecting document anomalies due to high variability across document types and countries, noise levels, lack of genuine reference documents, limited fraudulent data, and performance degradation with new fraud types, requiring a modular, adaptable, scalable, interpretable, and customized solution.
Innovation Solution
A method involving segmentation of document images into regions of interest (ROIs), applying transformations to generate transform-specific and region-specific features, and computing anomaly scores with modular thresholds to detect anomalies, allowing for local adaptability and interpretability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a global fraud detection model is trained across thousands of different document classes, then the model can handle diverse document types, but performance deteriorates when new fraud types are incorporated and model updates are delayed
Solution Approach 1:
The patent segments the document verification system into separate document class-specific models rather than a single global model. Each document class (e.g., passports, driver's licenses) has its own trained model, allowing rapid updates for new fraud types in specific classes without affecting other classes. This segmentation enables independent training on genuine documents for each class, maintaining high performance for new fraud types while preserving versatility across document types.
2Reliability
If comprehensive fraudulent data is used for training, then the model can detect known fraud types, but training is limited by lack of genuine reference documents for certain countries
Solution Approach 1:
The patent inverts the traditional training approach by training only on genuine documents rather than requiring fraudulent documents for training. The model learns the characteristics of authentic documents and automatically detects anomalies that deviate from this learned pattern. This inversion eliminates the need for comprehensive fraudulent data while enabling the model to adapt to new countries and document classes using only genuine reference documents from those classes.
3Device complexity
If a single global anomaly detection model is used, then the system is simple to implement, but the model cannot provide explanations for flagged anomalies making human verification inefficient
Solution Approach 1:
The patent segments the anomaly detection system into document class-specific models, each capable of providing detailed, interpretable explanations for flagged anomalies. This segmentation allows human verifiers to understand which specific characteristics of a document triggered the anomaly detection, significantly improving verification efficiency while maintaining reasonable system complexity through modular architecture.
Data Source
AI summary
Described are methods and systems for training a system for detecting anomalies in images of documents in a class of documents. A plurality of training document images of training documents in a class of documents are obtained. For each training document image, the training document image is segmented into a plurality of region of interest (ROI) images, each ROI image corresponding to a respective ROI of the training document. For each ROI image, a plurality of transformations are applied to the ROI image to generate respective transform-specific features for the ROI image and respective transform-specific anomaly scores from the transform-specific features. Based on the respective anomaly scores of the plurality of training document images, a transform-specific threshold is computed for each transformation to separate document images containing an anomaly from document images not containing an anomaly.