Neural Network Handwritten Payee Recognition on Scanned Checks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automated processes in banks are unable to detect handwritten payees on scanned checks, requiring users to click on the check image to view transaction details, which is inconvenient.
Innovation Solution
A system and method using a neural network classification model to identify handwritten characters on images by converting typed text to machine-encoded text, locating handwritten text, clustering non-white pixels, generating separate images for each character, classifying characters, and determining the accuracy through a secondary database search.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If automated processes use current detection methods, then processing speed is maintained, but handwritten payee detection accuracy is insufficient
Solution Approach 1:
The system segments the check image processing into distinct stages: OCR for typed text, neural network classification for handwritten characters, and database verification for payee identification. Each segment handles specific tasks with specialized algorithms, improving overall detection accuracy while maintaining manageable system complexity through modular architecture.
Solution Approach 2:
The patent introduces an intermediary neural network classification system between the initial OCR processing and the final payee identification. This intermediary layer specialized in handwritten character recognition bridges the gap between typed text detection and payee verification, enhancing accuracy without overwhelming system complexity.
2Ease of operation
If users manually click on check images to view transaction details, then detection accuracy is maintained, but user convenience deteriorates
Solution Approach 1:
The system implements self-service automation where the neural network classification model automatically detects and identifies handwritten payees without requiring user intervention. The system serves itself by processing handwritten characters through the classification pipeline and presenting results directly in the transaction register, eliminating the need for users to manually click on check images.
Solution Approach 2:
The patent performs preliminary automated detection of handwritten payees before the user views the transaction details. The neural network classification and database verification processes run in advance, preparing the payee information ready for display, so users receive fully processed information without needing to perform additional actions.
3Loss of information
If the system displays only transaction date and amount, then processing simplicity is maintained, but information completeness deteriorates
Solution Approach 1:
The system performs preliminary processing of handwritten payee information through neural network classification and database verification before final display. This preliminary action ensures that complete payee information is ready for immediate presentation alongside transaction date and amount, enhancing information completeness without compromising processing efficiency.
Solution Approach 2:
The patent replaces manual information gathering with automated neural network-based character recognition and database lookup. This substitution of mechanical/user-driven processes with automated intelligent systems enables complete payee information to be captured and displayed efficiently, preventing information loss while maintaining productivity.
Data Source
AI summary
A system for identifying handwritten characters on an image using a classification model that employs a neural network. The system includes a computer having a processor and a memory device that stores data and executable code that, when executed, causes the processor to read and convert typed text on the image to machine encoded text to identify locations of the typed text on the image; identify a location on the image that includes handwritten text based on the location of predetermined typed text on the image; identify clusters of non-white pixels in the image at the location having the handwritten text; generate an individual and separate cluster image for each identified cluster; classify each cluster image using machine learning and at least one neural network to determine the likelihood that the cluster is a certain character; and determine the accuracy of the characters by comparing to a secondary database.