Character Recognition Apparatus Fixed-Pitch Font Space Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Character recognition technologies fail to accurately distinguish between original and erroneous space characters in documents with fixed-pitch fonts, leading to incorrect word separation and inadequate correction processing, especially in Japanese business forms and documents with fixed-pitch fonts like MS Gothic.
Innovation Solution
A character recognition apparatus that forms circumscribed rectangles around character strings, determines if the font is fixed-pitch by analyzing the distance between center lines of adjacent rectangles, identifies excess space characters based on their width, and deletes them to correct the character string, using a computer program to execute these processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If space characters are recognized based on simple distance thresholds between characters, then the recognition process is simple and fast, but erroneous space characters are incorrectly identified and deleted
Solution Approach 1:
The patent changes the parameter for space character identification from a simple fixed threshold to a dynamic threshold based on character width ratios. By calculating the ratio of space character width to preceding character width and comparing it against predetermined ranges, the system adapts to different font characteristics and document types, thereby improving identification accuracy without significantly impacting processing speed.
Solution Approach 2:
The patent implements feedback by using the recognized character string and its layout information to validate and adjust space character identification. The system references the original document image and character positions to verify whether identified space characters are genuine, allowing for iterative refinement and correction of recognition results.
2Device complexity
If a fixed threshold is used for space character width to distinguish original and erroneous spaces, then the processing is simple, but it fails to handle variable font widths and document types
Solution Approach 1:
The patent transforms the fixed threshold parameter into a relative parameter based on character width ratios. By calculating the ratio of space width to preceding character width and using predetermined ranges rather than absolute values, the system automatically adapts to different font sizes, styles, and document types, significantly improving versatility while maintaining reasonable processing complexity.
Solution Approach 2:
The patent introduces dynamic adjustment of the space character identification criteria based on the actual character widths detected in the document. The system dynamically calculates width ratios and adjusts its recognition thresholds accordingly, allowing it to adapt to varying document characteristics without requiring manual reconfiguration.
3Measurement precision
If language analysis correction is performed on the character string, then recognition accuracy improves, but processing time increases and erroneous spaces disrupt the correction process
Solution Approach 1:
The patent performs preliminary action by identifying and removing erroneous space characters before conducting language analysis correction. By first filtering out false space characters based on width ratio criteria, the system prepares a cleaned character string that is more suitable for subsequent correction processing, thereby improving efficiency and accuracy.
Solution Approach 2:
The patent extracts and removes the harmful element (erroneous space characters) from the character string before performing the main correction process. By separating and eliminating these disruptive elements, the system prevents them from interfering with language analysis and morphological correction, thereby reducing processing time and improving overall accuracy.
Data Source
AI summary
The character recognition apparatus recognizes characters from a read document original to correct a character string as a character recognition result in a word unit with a space character as a separator. The character recognition apparatus includes a circumscribed rectangle formation portion which forms a circumscribed rectangle for each recognized alphabet character string, a fixed-pitch font determination portion which determines whether or not a font is a fixed-pitch font based on a distance between center lines in a width direction of adjacent circumscribed rectangles, a portion for determining an excess space character which determines, in the case of a fixed-pitch font, that the space character is an excess based on that a width of a space character in the character string is narrower than a predetermined width, and a portion for deleting the space character determined as an excess from the character string.