基于结构特征的汉字识别

Chinese character recognition based on structural features

提出一种基于文字结构特征的神经网络手写汉字识别策略 ,根据所提取的文字笔画方向、基本轮廓和交叉点等特征 ,采用基于自组织神经网络的模式聚类该方法完成正规手写文字的识别 .该方法提取的笔画轮廓十分准确有效 ,对手写汉字的约束少 ,可识别的汉字数量大 。

A structural feature based handwritten Chinese character recognition method using self organizing neural network is presented. Using this method, regular handwritten Chinese characters is recognized by extraction of certain features such as stroke direction, basic skeleton, and fork segments as well as with neural network based pattern classification. It can extract accurate stroke profiles and recognize most of handwritten Chinese characters efficiently without any specific restriction to the characters as witnessed by a simulation test.