计算机应用 ›› 2010, Vol. 30 ›› Issue (3): 768-771.

• 人工智能 • 上一篇    下一篇

基于特征反馈的手写体汉字识别系统研究

王建平1,王二帅2   

  1. 1. 合肥工业大学
    2. 安徽合肥合肥工业大学南区电气与自动化工程学院07级18班
  • 收稿日期:2009-09-08 修回日期:2009-11-02 发布日期:2010-03-14 出版日期:2010-03-01
  • 通讯作者: 王二帅

Feature feedback-based system for recognition of handwritten Chinese characters

1,   

  • Received:2009-09-08 Revised:2009-11-02 Online:2010-03-14 Published:2010-03-01

摘要: 针对传统基于开环的汉字识别系统不完全符合人类识字过程的问题,构建了一种具有反馈结构的仿人智能识别系统。该系统根据待识别汉字的多模态定性识别结果来选择最佳的首轮识别方案,在完成识别之后,提取广义识字误差对候选字进行可信度判断和反馈校正。设计了3种广义识字误差,通过对这3种广义识字误差的类型和数值进行定性与定量相结合的分析,建立了识别结果的可信度评价指标体系和反馈校正决策机制。仿真实验结果验证了方法的可行性。

关键词: 特征反馈, 手写体汉字识别, 广义识字误差, 可信度, 反馈校正

Abstract: Due to the difficulty of conforming to human recognition process with traditional open loop recognition system, a simulated intelligent recognition system with feedback structure was constructed. Choosing the best initial recognition method based on the multimodal and qualitative recognition result, the system extracted general characters recognition error to judge the credibility and feedback correction after the initial recognition. Three kinds of general characters recognition error were designed according to the feedback result. The credibility evaluation index system and feedback correction decision-making mechanism of the recognition result were established by qualitative and quantitative analysis of three kinds of general characters recognition error. And the analytical method of the recognition error was given. The experimental results prove that the presented method is effective.

Key words: feature feedback, recognition of handwritten Chinese characters, general characters recognition error, credibility, feedback correction