Ledger Recognition System Using Multiple OCR Algorithms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current ledger recognition systems have low accuracy for handwritten characters, leading to significant manual correction burdens in financial institutions, as they rely on a single OCR recognition program and struggle to differentiate between similar characters like katakana and Chinese characters.
Innovation Solution
A ledger recognition system utilizing a headquarter server connected via a public telecommunication network with multiple OCR recognition programs having different algorithms to recognize and correct handwritten characters, including number determination and user name correction processing, by comparing results from multiple OCR programs and utilizing characteristic analysis for ambiguous numbers and syllabary conversion for katakana characters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single OCR recognition program is used for handwritten character recognition, then the device complexity is low, but the recognition accuracy is insufficient
Solution Approach 1:
The patent combines multiple OCR recognition programs with different algorithms into a unified recognition system. The server executes several OCR programs simultaneously on the same handwritten character image, merging their recognition results through comparison and validation logic to achieve higher accuracy than any single program could provide alone.
Solution Approach 2:
The recognition system uses a composite approach by integrating multiple OCR algorithms with different recognition strategies. Each OCR program contributes its unique algorithmic characteristics, and the system combines these diverse recognition methods into a unified decision-making process that leverages the strengths of each individual program.
2Measurement precision
If multiple OCR recognition programs are used to improve recognition accuracy, then the recognition accuracy is enhanced, but the processing time increases
Solution Approach 1:
The system does not require all OCR programs to complete full processing before producing results. Instead, it uses partial results from multiple programs and applies validation logic to determine the correct recognition early in the process, avoiding the need to wait for all programs to finish their complete execution cycles.
Solution Approach 2:
The system implements feedback mechanisms where recognition results from different OCR programs are compared and validated against each other. When results agree or when validation criteria are met, the system can confidently determine the correct character without requiring additional processing time from all programs.
3Measurement precision
If manual correction is performed for unrecognized handwritten characters, then the recognition accuracy is improved, but the operational burden increases
Solution Approach 1:
The system performs self-correction by automatically comparing recognition results from multiple OCR programs and using validation logic to identify and correct errors. The server autonomously determines the correct character recognition without requiring manual intervention, making the system self-sufficient in handling recognition challenges.
Solution Approach 2:
The system uses feedback from multiple OCR programs to automatically correct recognition errors. By comparing results and applying validation rules, the system identifies discrepancies and corrects them automatically, reducing the need for manual correction while maintaining high accuracy.
Data Source
AI summary
Provided is a ledger recognition system which can enhance recognition accuracy of a handwritten character filled out by a user thus capable of reducing a manual work in a correction operation. A ledger recognition system includes: a headquarter server configured to recognize handwritten characters described in a ledger by a user; a system terminal including an image scanner for reading the handwritten characters filled out in the ledger by the user; and a public telecommunication network which allows the headquarter server and the system terminal to be communicably connected with each other. The headquarter server includes a handwritten character recognition unit where the handwritten character recognition unit receives the image data of the ledger read by the image scanner from the system terminal, recognizes the handwritten characters written by the user in the image data of the received ledger in accordance with at least two types of OCR recognition programs having different algorithms, determines the handwritten characters described in the ledger with respect to a part of the handwritten characters where recognition results in accordance with the OCR recognition programs agree with each other, and sets a part of the handwritten characters where the recognition results by the OCR recognition programs do not agree with each other as an object of correction processing.