Check Image Authentication Using Fallback Handwriting Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems face challenges in authenticating checks for deposit by comparing reference images and handwriting, particularly when reference images are unavailable, leading to potential fraud and the need for robust security controls.
Innovation Solution
A system utilizing machine learning models to analyze captured check images for image and handwriting similarity, including color analysis, to determine authenticity, and interface with a server for handwriting profiles, enabling secure deposit verification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If reference image comparison is used for check authentication, then authentication accuracy is improved, but the system fails when reference images are unavailable
Solution Approach 1:
The patent introduces handwriting analysis as an intermediary method that bridges the gap when reference image comparison cannot be performed. The system uses machine learning models to analyze handwriting characteristics and compare them against stored handwriting profiles, enabling authentication to proceed even without reference images of the check itself.
Solution Approach 2:
The system performs preliminary actions by storing handwriting profiles during account setup and reference image comparisons when available. These preliminary preparations enable the system to fallback to alternative authentication methods when primary methods fail, ensuring continuous operation.
2Reliability
If multiple machine learning models are invoked for comprehensive analysis, then fraud detection capability is improved, but processing time and computational resources increase
Solution Approach 1:
The patent segments the authentication process into distinct analysis stages: image quality assessment, reference image comparison (when available), handwriting analysis, and color analysis. Each stage uses specialized machine learning models that process specific aspects independently, allowing for optimized processing and early termination when fraud is detected at any stage.
Solution Approach 2:
The system performs partial analysis by selectively invoking machine learning models based on available data and risk assessment. When reference images are unavailable, the system focuses on handwriting and color analysis rather than performing all possible checks, balancing thoroughness with processing efficiency.
3Reliability
If security controls are initiated after predetermined time without approval, then system security is improved, but user convenience deteriorates
Solution Approach 1:
The patent implements periodic security checks by initiating security controls after a predetermined time without approval. This timed intervention balances security requirements with user convenience, allowing legitimate transactions to proceed smoothly while detecting and blocking fraudulent attempts that show unusual delays or patterns.
Data Source
AI summary
A system for interfacing with a server over a network to improve efficiency of image processing using multiple machine learning models. The system receives a captured image that is captured via a mobile application and interfaces with the server, over the network, and via an application program interface (API) to determine that a reference image, required for use of the first machine learning model, has not been provided for the captured image. The system further reserves an execution of the second machine learning model for the captured image until the first machine learning model is insufficient for the captured image and performs the execution of the second machine learning model for the captured image based on reserving the execution of the second machine learning model until the first machine learning model is insufficient for the captured image and based on determining that the reference image is not provided for the captured image.