基于手写体汉字的成绩单自动识别系统

A System for Automatic Recognition of Hand Written Characters on Scorecards

本文针对基于手写体汉字的成绩单自动识别系统的实际需要 ,基于目前脱机手写体汉字识别技术的一些研究成果 ,提出了一种新的基于聚类方法和多个分类器的线性集成的综合方法 ,并且采用分级判决策略 ,进一步降低误识率 ,实验结果表明 ,本系统具有较高的识别率和较低的误识率 ,并且具有较好的扩展性 。

This paper proposes a new method of recognizing hand written Chinese characters which combines clustering and linear integration of multiple classifications based on recent researches in this field. It adopts the strategy of hierarchical judgements and decreases the rate of erroneous judgement. Experiments show that this system is characterized with higher rate of correct judgement and lower rate of erroneous judgement, and with expandability and feasibility.