Character Rectangle Correction for Small Handwritten OCR Marks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing character recognition technologies often misrecognize small-size handwritten characters such as commas and punctuation marks as larger characters, leading to inaccurate OCR results.
Innovation Solution
An image processing apparatus that extracts character-sequence and one-character rectangles, corrects specific rectangles based on predetermined conditions, and performs character recognition using the corrected rectangles to improve accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional OCR processing is applied to character sequences including small-size handwritten characters, then processing speed is maintained, but character recognition accuracy deteriorates
Solution Approach 1:
The patent segments the character recognition process into multiple stages: first extracting character-sequence rectangles, then extracting one-character rectangles, and finally correcting specific rectangles based on size and position conditions. This segmentation allows the system to handle different character types differently, improving accuracy for small-size characters without compromising overall processing efficiency.
Solution Approach 2:
The patent applies local quality by treating small-size handwritten characters differently from regular-sized characters. The correction processing unit specifically targets one-character rectangles that satisfy predetermined conditions (indicating small-size characters) and applies correction only to these local regions, leaving other characters unchanged. This localized approach improves recognition accuracy for problematic characters while maintaining overall processing speed.
2Reliability
If small-size handwritten characters are processed without correction, then processing time is reduced, but recognition errors increase
Solution Approach 1:
The patent performs preliminary extraction of character-sequence rectangles and one-character rectangles before the actual recognition processing. The correction processing unit identifies and corrects specific rectangles in advance based on their size and position characteristics. This preliminary action ensures that when the recognition processing unit operates, the data is already optimized, reducing the need for time-consuming corrections during the main processing phase.
3Measurement precision
If all one-character rectangles are processed uniformly, then processing simplicity is maintained, but recognition accuracy for small characters deteriorates
Solution Approach 1:
The patent implements local quality by differentiating the processing treatment based on character characteristics. The correction processing unit applies predetermined conditions to identify small-size handwritten characters (one-character rectangles with specific size and position properties) and applies correction only to these local cases. This selective approach maintains processing simplicity for the majority of characters while improving accuracy for the minority of problematic small characters.
Data Source
AI summary
An image processing apparatus includes an acquisition processing unit that acquires character image data, a character sequence extraction processing unit that extracts a character-sequence rectangle corresponding to a character sequence composed of a plurality of characters from the character image data, a one-character extraction processing unit that extracts a plurality of one-character rectangles corresponding to the plurality of characters, respectively, from the character image data, a correction processing unit that corrects a specific one-character rectangle when the plurality of one-character rectangles include the specific one-character rectangle, at least one of a position and a size of the specific one-character rectangle satisfying a predetermined condition, and a recognition processing unit that executes character recognition processing on the character sequence using the specific one-character rectangle corrected by the correction processing unit.