Mobile Check Image Profiling for Adaptive Fraud Scoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional check fraud detection systems face issues with static CIRs leading to compromised fraud detection confidence over time, image quality affecting data extraction, and inaccuracy due to low resolution, skewed images, redactions, and non-US bank checks, lacking image comparison and overlay controls.
Innovation Solution
A method involving a hardware processor to receive incoming check images, extract features, compare them with stored profiles, and develop fraud scores using multiple detectors, incorporating OCR and neural networks for improved accuracy across various check types.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional static CIRs are used for fraud detection, then the system is simple to implement, but fraud detection confidence is compromised over time
Solution Approach 1:
The patent implements dynamic CIRs that automatically update with new check images over time. The system transitions from static profiles to dynamic profiles that adapt to changing check formats and patterns, maintaining fraud detection confidence through continuous learning while preserving system simplicity through automated updates.
Solution Approach 2:
The system incorporates feedback mechanisms where fraud detection results and new check images continuously update the CIR database. This feedback loop allows the system to learn from new data, improving detection accuracy over time while maintaining operational simplicity through automated processes.
2Ease of operation
If image processing is performed on low resolution or skewed images, then mobile deposit accessibility is improved, but data extraction accuracy deteriorates
Solution Approach 1:
The system performs preliminary image quality assessment and preprocessing before fraud detection. Images are evaluated for resolution, skew, and quality metrics, then preprocessed through correction algorithms to improve extraction accuracy before analysis, enabling mobile deposit accessibility without sacrificing precision.
Solution Approach 2:
The patent applies parameter changes through image preprocessing techniques that adjust resolution, correct skew angles, and optimize image quality metrics. These parameter transformations enable accurate data extraction from mobile-captured images while maintaining accessibility for users with various device capabilities.
3Measurement precision
If multiple detectors and neural networks are implemented for fraud detection, then detection accuracy is improved, but system complexity increases
Solution Approach 1:
The patent segments the fraud detection system into multiple specialized detectors, each focusing on specific fraud indicators (e.g., amount field analysis, signature verification, check pattern recognition). This segmentation improves overall detection accuracy by dividing complex analysis into manageable specialized components while organizing system complexity through modular architecture.
Solution Approach 2:
The system implements universal neural network components that perform multiple functions across different detector types. These multi-functional elements reduce overall system complexity by reusing proven algorithms and data structures across multiple detection scenarios, improving accuracy without proportionally increasing complexity.
4Reliability
If strict image quality requirements are enforced, then data extraction reliability is improved, but acceptance of diverse check formats deteriorates
Solution Approach 1:
The patent applies local quality standards to different regions and aspects of check images rather than enforcing uniform global requirements. Critical areas like amount fields and signatures receive stricter quality assessment, while other regions tolerate more variation, enabling reliable extraction from diverse check formats including non-US banks while maintaining high reliability for critical data.
Data Source
AI summary
A method comprising using at least one hardware processor to: receiving a new incoming check image associated with an account from a mobile deposit application; extracting from the incoming check its features, wherein the features are associated with a plurality of detectors; comparing the features with corresponding features associated with profile check images stored in a CIR; developing one or more fraud scores based on the comparisons, for each of the plurality of detectors; and returning the fraud scores to the mobile deposit application.