Superimposed Handwriting Recognition via Concurrent Expert Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing on-line handwriting recognition systems struggle with recognizing superimposed handwriting, where characters are written one over another without explicit separation, especially in languages like Japanese or Chinese, and in cursive writing.
Innovation Solution
The system performs concurrent segmentation, recognition, and interpretation using a collaborative approach between a segmentation expert, recognition expert, and language expert through dynamic programming, allowing for the recognition of superimposed handwriting and cursive writing without explicit character separation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multi-stroke recognition is used to recognize natural handwriting in languages like Japanese or Chinese, then the system can handle complex characters, but character segmentation becomes difficult when characters are written continuously without explicit separation
Solution Approach 1:
The patent applies segmentation by dividing the continuous handwriting input into individual character units through automated detection of character boundaries. The system segments superimposed handwriting by analyzing stroke patterns and spatial relationships, breaking down the continuous input into discrete characters without requiring explicit user separation gestures.
Solution Approach 2:
The patent transitions from analyzing handwriting in a single temporal dimension to incorporating spatial dimension analysis. By examining the two-dimensional spatial relationships between strokes and using superimposition techniques, the system can distinguish character boundaries even when characters are written continuously, adding a spatial dimension to the recognition process.
2Ease of manufacture
If sequential processing stages (preprocessing, segmentation, recognition, interpretation) are used, then each stage can be optimized independently, but the overall recognition accuracy for superimposed handwriting remains insufficient
Solution Approach 1:
The patent merges the traditionally sequential processing stages into a concurrent, integrated system. Multiple experts (segmentation expert, recognition expert, language expert) operate simultaneously and interactively, sharing information and refining results in real-time. This concurrent processing allows the system to achieve higher recognition accuracy for superimposed handwriting by leveraging the combined strengths of all processing stages working together.
Solution Approach 2:
The patent implements feedback mechanisms where the results from one expert stage are fed back to influence the work of other experts. The segmentation expert uses feedback from recognition and language experts to refine character boundaries, while recognition accuracy is improved through feedback from language models. This iterative feedback loop continuously improves recognition accuracy for challenging superimposed handwriting.
Data Source
Figure 1
Figure 2
Figure 3A~3B
AI summary
A system and method that is able to recognize a user's natural superimposed handwriting without any explicit separation between characters. The system and method is able to process single- stroke and multi-stroke characters. It can also process cursive handwriting. Further, the method and system can determine the boundaries of input words either by the use of a specific user input gesture or by detecting the word boundaries based on language characteristics and properties. The system and method analyzes the handwriting input through the processes of segmentation, character recognition, and language modeling. These three processes occur concurrently through the use of dynamic programming.