Path Score Calculation for Handwriting Recognition Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current handwriting recognition systems face challenges in achieving high accuracy, particularly with cursive handwriting, often resulting in less than 60% accuracy, which does not meet customer requirements due to variations in human handwriting styles.
Innovation Solution
A machine learning method that calculates path scores by detecting connected regions, determining segmentation positions, obtaining recognition results and scores, and using the path with the highest score to interpret handwritten characters, enhancing baseline estimation and segmentation through clustering and contour analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional handwriting recognition methods are used, then the system is simple and easy to implement, but the recognition accuracy is low (no more than 60% for cursive handwriting)
Solution Approach 1:
The patent segments the handwriting recognition process into multiple paths, each representing a different interpretation of the handwritten input. The system divides the recognition space into multiple possible paths and evaluates each path independently, then selects the best path based on scoring. This segmentation allows the system to explore multiple interpretations simultaneously, improving accuracy for cursive handwriting where character boundaries are ambiguous.
Solution Approach 2:
The patent introduces path scores as a new parameter to evaluate different recognition paths. By calculating and comparing scores across multiple paths, the system can objectively determine the most likely interpretation. The scoring mechanism transforms the recognition problem from a single-threshold decision into a multi-parameter evaluation, enabling more nuanced and accurate recognition decisions.
2Measurement precision
If multiple recognition paths are evaluated to improve accuracy, then recognition performance improves, but the calculation complexity and processing time increase
Solution Approach 1:
The patent performs preliminary actions by pre-defining the structure of recognition paths and pre-calculating segment scores before final path evaluation. The system prepares the recognition space by identifying potential paths and their constituent segments in advance, which reduces the computational burden during the actual recognition process. This preliminary structuring allows for more efficient evaluation when multiple paths need to be compared.
3Measurement precision
If path scoring is used to identify the most accurate path, then interpretation accuracy improves, but the computational resources required increase
Solution Approach 1:
The patent applies partial action by evaluating only the necessary components of each path rather than exhaustively analyzing all possible interpretations. The system calculates scores for segments and paths selectively, focusing computational resources on the most promising recognition paths. This approach achieves high accuracy without requiring proportional increases in computational energy by being strategic about which paths and segments receive full evaluation.
Data Source
AI summary
Disclosed herein is a method that improves the performance of handwriting recognition by calculating path scores so as to identify the path with the highest score as the basis for interpreting handwritten characters. Specifically, the method comprises the following steps: detecting connected regions in an input image comprising handwritten characters; determining a plurality of segmentation positions of the input image; obtaining a plurality of recognition results for each segment of each path in the input image, wherein each recognition result represents a character candidate for the segment and each path comprises one or more segments; obtaining a plurality of scores corresponding to the recognition results; calculating scores for each path in the input image based on segment lengths and the scores corresponding to the recognition results; and using the path with the highest score to interpret the handwritten characters in the input image.