OCR Payee Identifier Matching for Bill Payment Automation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing electronic bill payment systems require users to manually parse and enter information from bills, leading to errors and frustration, as they often rely on manual input of payee and account details from either electronic or printed formats.
Innovation Solution
A system and method utilizing optical character recognition (OCR) to auto-populate electronic transaction processes by scanning payee identifiers from digital bill documents, comparing them with stored identifiers, and selecting the best match based on scoring, which includes features like postal code, PO Box, URL, and logo detection without cloud-based deep neural networks, using techniques like k-means clustering and head/tails breaks for accurate identification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual parsing and entry of bill information is used, then users can complete transactions, but error rate increases and user frustration increases
Solution Approach 1:
The system enables self-service by allowing the bill document itself to provide the data automatically. OCR technology extracts payee name, account number, and other identifiers directly from the bill image, eliminating the need for manual transcription and reducing errors caused by human input.
Solution Approach 2:
The patent replaces the mechanical manual typing process with automated optical character recognition and image processing. The system uses OCR to convert visual text from the bill into machine-readable data, and employs pattern matching algorithms to automatically populate transaction fields without user intervention.
2Measurement precision
If cloud-based deep neural networks are used for OCR, then identification accuracy improves, but data privacy is compromised and processing time increases
Solution Approach 1:
The patent segments the OCR processing into distinct functional components: preprocessing (deskewing, noise reduction), character recognition, and post-processing (validation, disambiguation). This modular approach allows accurate recognition to be achieved through coordinated simple operations rather than requiring complex cloud-based neural networks.
Solution Approach 2:
The system introduces an intermediary layer of rule-based validation and disambiguation logic between the raw OCR output and the final transaction data. This intermediary process uses known patterns (e.g., account number formats, postal code structures) to correct and verify OCR results, achieving high accuracy without needing to send data to the cloud.
3Measurement precision
If multiple OCR scanning methods are used, then identification accuracy improves, but processing complexity increases
Solution Approach 1:
The patent implements dynamic processing where the system adapts its OCR approach based on the input characteristics. It applies preprocessing techniques such as deskewing and noise reduction selectively, and uses iterative refinement where initial OCR results are validated and corrected based on known data patterns, achieving high accuracy without requiring all possible processing methods simultaneously.
Data Source
AI summary
A system and method for auto-populating an electronic transaction process is provided. The system comprises at least one processor, and a memory storing instructions which when executed by the at least one processor configure the processor to obtain a scanned payee identifier from an optical character recognition scan of a digital bill document, compare the scanned payee identifier with a set of stored payee identifiers to obtain at least one first identifier match, determine a score for each of the at least one identifier match, and select the stored payee identifier associated with a highest score. The method comprises obtaining a scanned payee identifier from an optical character recognition scan of a digital bill document, comparing the scanned payee identifier with a set of stored payee identifiers to obtain at least one first identifier match, determining a score for each of the at least one identifier match, and selecting the stored payee identifier associated with a highest score.