Handwritten Text Segmentation via Stroke Thresholds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current text line segmentation technologies for handwritten input, primarily based on horizontal and vertical projection histograms, suffer from poor accuracy, which negatively impacts character recognition systems.
Innovation Solution
A method for handwritten text recognition that involves acquiring track points with abscissa, writing time, and state values, dividing them into strokes based on writing state values, calculating segmentation thresholds, and determining segmentation points to segment text accurately, using statistical information and thresholds to differentiate between text lines.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional projection histogram methods are used for text line segmentation, then the implementation is simple, but the segmentation accuracy is poor
Solution Approach 1:
The patent changes the parameter basis for segmentation from projection histograms to stroke-level features including abscissa differences, writing time intervals, and writing state values. This parameter transformation enables more accurate text line segmentation by directly utilizing the structural and temporal characteristics of handwriting strokes.
Solution Approach 2:
The patent segments the handwritten text into individual strokes first, then performs text line segmentation based on stroke characteristics. This two-level segmentation approach (stroke segmentation followed by text line segmentation) improves accuracy by analyzing text structure at multiple granularities rather than treating the entire text as a single unit.
2Adaptability or versatility
If fixed segmentation thresholds are used, then the method is simple to implement, but it cannot adapt to different writing styles and accuracyle
Solution Approach 1:
The patent introduces dynamic segmentation thresholds that are calculated based on statistical analysis of the input handwriting data. The thresholds adapt to different writing styles by computing mean and variance of stroke characteristics (abscissa differences, writing time intervals) from the actual input, making the segmentation process flexible and adaptive rather than rigid and fixed.
Solution Approach 2:
The patent implements a feedback mechanism where the segmentation process uses statistical information from the analyzed handwriting data to adjust and optimize segmentation thresholds. By calculating mean and variance from the input strokes and using these statistics to determine adaptive thresholds, the system continuously optimizes its segmentation performance based on the characteristics of the specific handwriting being processed.
Data Source
AI summary
The present disclosure relates to a handwritten text recognition method, including: acquiring an information sequence including a plurality of track points of handwritten text, wherein information on each track point comprises its abscissa, writing time and writing state value; dividing the plurality of track points into a plurality of strokes according to the writing state value of each track point, the writing state value including a first value representative of stroke pen-up and a second value representative of stroke pen-down, respectively; calculating a first segmentation threshold of the handwritten text; determining a first text segmentation point according to a result of comparison between an absolute value of a difference between abscissas of a start track point of one stroke and an end track point of its previous stroke and the first segmentation threshold; and performing text segmentation according to the first text segmentation point to obtain a text segmentation result.