Removing Connective Strokes in Cursive Handwriting Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Handwritten input in cursive form, such as logograms or alphabet-based text, often includes connective strokes that impede recognition, as existing technologies fail to accurately identify and remove these strokes, leading to unreliable and inaccurate input recognition.
Innovation Solution
An apparatus comprising a written input device, processor, and memory that identifies connective points in a contiguous input based on a connective score exceeding a threshold, and removes the connective strokes between these points, utilizing a method and computer program product to enhance recognition accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If connective strokes are retained in cursive handwritten input, then the natural flow and continuity of writing is preserved, but recognition accuracy deteriorates due to ambiguous character boundaries
Solution Approach 1:
The patent segments the continuous cursive handwritten input by identifying and removing connective strokes that link separate characters. This segmentation process divides the contiguous input into distinct character units, enabling accurate recognition while preserving the natural cursive writing style. The system calculates connective scores for potential connective points and removes strokes where the connective score exceeds a threshold, effectively segmenting the input without requiring the user to change their writing manner.
2Measurement precision
If existing technologies attempt to identify connective strokes, then stroke removal is possible, but accuracy deteriorates due to unreliable identification methods
Solution Approach 1:
The patent employs a feedback mechanism by calculating a connective score for each potential connective point based on multiple features including stroke curvature, pressure variations, and directional changes. The system uses this score as feedback to determine whether a point is indeed a connective point (when score exceeds threshold) or a character boundary point (when score is below threshold), thereby accurately identifying connective strokes for removal and improving overall recognition reliability.
Solution Approach 2:
The patent changes multiple parameters simultaneously to improve connective point identification accuracy, including stroke curvature, pressure variations, directional changes, and connective scores. By analyzing these parameters together and using their combined information to calculate connective scores, the system achieves more accurate identification of connective strokes compared to existing technologies that rely on single-parameter methods.
3Measurement precision
If connective strokes are removed from contiguous written input, then character recognition accuracy improves, but the processing time increases
Solution Approach 1:
The patent performs preliminary action by pre-calculating and storing connective scores for all potential connective points in the handwritten input before final character recognition. This preliminary calculation of connective scores based on stroke features allows the system to quickly identify and remove connective strokes during the recognition process, reducing overall processing time while maintaining high character recognition accuracy.
Data Source
AI summary
For removing connective strokes, code identifies two connective points in a contiguous written input from a written input device. Each connective point is identified in response to a connective score for the connective point exceeding a connective threshold. The code further removes a connective stroke between the two connective points from the contiguous written input.