Information Processing Apparatus for Non-Fixed Format Image Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing information processing systems are inefficient in extracting information from images with non-fixed formats, as they require manual specification of character recognition areas, leading to time-consuming processing.
Innovation Solution
An information processing apparatus that performs area analysis and character recognition by defining keywords and value conditions, determining the order of specifying areas based on these rules, and performing character recognition processing to efficiently extract information from images with unknown formats.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If character recognition processing is performed on the entire image to handle non-fixed formats, then information extraction capability is improved, but processing time increases significantly
Solution Approach 1:
The patent divides the image into multiple candidate areas based on coordinate information from similar forms, then performs character recognition only on these segmented regions rather than the entire image. This reduces processing time while maintaining the ability to handle non-fixed formats by focusing computational resources on relevant areas only.
Solution Approach 2:
The patent performs preliminary area analysis to identify candidate regions before executing character recognition processing. By pre-determining which areas are likely to contain target information based on coordinate data from similar forms, the system avoids unnecessary processing of irrelevant regions, thus reducing overall processing time while preserving adaptability.
2Measurement precision
If specific character recognition processing is performed on the entire image, then recognition accuracy is improved, but it requires double processing time for re-recognition targets
Solution Approach 1:
The patent extracts only the candidate areas that are likely to contain target information based on coordinate information from similar forms, and performs character recognition exclusively on these extracted regions. This eliminates the need for double processing of entire images while maintaining recognition accuracy by focusing on relevant areas.
Solution Approach 2:
Instead of performing character recognition on the entire image, the patent applies partial action by processing only the identified candidate areas. This partial processing approach achieves sufficient recognition accuracy for the target information without the time cost of processing the complete image, especially avoiding redundant re-recognition processing.
3Measurement precision
If manual specification of character recognition areas is used for non-fixed formats, then processing accuracy is improved, but operation complexity increases
Solution Approach 1:
The patent enables the system to automatically determine candidate areas using coordinate information from similar forms, eliminating the need for manual area specification. This self-service approach maintains processing accuracy by intelligently identifying relevant regions while significantly reducing operational complexity and user burden.
Solution Approach 2:
The patent utilizes coordinate parameters from similar forms to dynamically determine the areas for character recognition in non-fixed formats. By changing from manual coordinate specification to automatic parameter-based area determination, the system achieves both high processing accuracy and ease of operation.
Data Source
AI summary
An information processing apparatus extracts an area by performing an area analysis on an image, acquires a rule that defines a keyword and conditions of a value corresponding to the keyword, determines an order of specifying an area including the keyword and an area including the value corresponding to the keyword based on the acquired rule, firstly specifies the area including the keyword or the area including the value corresponding to the keyword from among the extracted area in accordance with the determined order, performs character recognition processing on the specified area, and secondly specifies the corresponding another area based on the acquired rule and the first specified area.