Related Experiment Video
Updated: May 29, 2026
05:58
Digital Handwriting Analysis of Characters in Chinese Patients with Mild Cognitive Impairment
Published on: March 11, 2021
Text-line extraction in handwritten Chinese documents based on an energy minimization framework
1Qualcomm Korea R&D Center, Seoul, Korea. hikoo@ispl.snu.ac.kr
Summary
This study introduces a novel cost function for text-line extraction in handwritten documents, significantly improving accuracy. The new method enhances handwritten text recognition by addressing challenges like varying text orientation and line interference.
Area of Science:
- Computer Science
- Artificial Intelligence
- Document Image Analysis
Background:
- Text-line extraction from unconstrained handwritten documents is difficult due to variations in character scale, text orientation, and line interference.
- Existing methods struggle to accurately segment text lines in complex handwritten documents.
Purpose of the Study:
- To develop a new cost function for improved text-line extraction in handwritten documents.
- To address challenges of nonuniform character scale, varying text orientation, and text line interference.
Main Methods:
- Proposed a novel cost function incorporating text line interactions and curvilinearity.
- Introduced normalized measures based on estimated line spacing.
- Developed an optimization method tailored to the new cost function's properties.
Main Results:
- Achieved a 99.52% detection rate on a database of 853 handwritten Chinese document images.
- Attained a low error rate of 0.32%.
- Demonstrated superior performance compared to conventional text-line extraction methods.
Conclusions:
- The proposed cost function and optimization method effectively improve text-line extraction accuracy in challenging handwritten documents.
- The method shows significant advantages over traditional approaches for handwritten Chinese document analysis.