Rounding Reliability Detection in Image Processing Apparatus
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing techniques fail to reliably determine the accuracy of rounding processing on numerical values obtained through character recognition, leading to potential errors in verification results due to differing rounding methods among users or organizations.
Innovation Solution
An image processing apparatus and method that performs character recognition, acquires a learning result for rounding processing from a storage unit, and determines the reliability of the recognition based on the difference between calculated and rounded numerical values, using a learning database to associate rounding methods with user or organizational identifiers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If character recognition processing is performed on numerical values from scanned forms, then extraction of numerical data is achieved, but reliability of verification results deteriorates due to unintended rounding processing
Solution Approach 1:
The system performs preliminary rounding processing on the calculated verification values before comparing them with the extracted numerical values. By applying rounding in advance based on learned rounding methods, the system prevents rounding discrepancies from causing verification failures, thus maintaining both extraction accuracy and verification reliability
Solution Approach 2:
The system learns rounding methods from previously processed forms and uses this learned information to adjust the verification process. The feedback loop allows the system to adapt to organization-specific rounding conventions, improving verification reliability without compromising the accuracy of numerical extraction
2Reliability
If rounding processing is applied to numerical values, then verification accuracy may be improved, but device complexity increases due to need to learn and store rounding methods
Solution Approach 1:
The system automatically learns rounding methods from the forms it processes without requiring manual configuration. By extracting rounding patterns directly from the data and storing them in a learning database, the system eliminates the need for complex manual setup and maintenance of rounding method configurations
Solution Approach 2:
The learning database acts as an intermediary between the character recognition process and the verification process. It stores learned rounding methods and provides them to the verification module, simplifying the overall system architecture by centralizing rounding method management in a dedicated storage component
Data Source
AI summary
An image processing apparatus that can determine reliability of rounding processing on numerical values obtained by character recognition processing. The image processing apparatus performs character recognition of a plurality of numerical values from image data, acquires a learning result indicating a rounding processing method from a storage unit that stores the learning result, and determines reliability of the character recognition on the basis of a difference between a calculation result of the plurality of numerical values and a calculation result of the plurality of numerical values in a case where the rounding processing indicated by the acquired learning result is applied.