构造化线段特征矩阵的手写汉字识别

HANDWRITTED CHINESE CHARACTERS RECOGNITION USING STRUCTURED STROKE FEATURE MATRIX

本文介绍了一种利用线段特征矩阵进行匹配的手写汉字识别方法。对输入文字图像测定其笔划宽度,抽取四个方向子图像。然后,利用文字图像重心分割图像成若干区域,按分割的区域,求各子图像区域的线段特征矩阵,与样本字库比较识别,进行手写汉字识别分类。实验表明本方法是有效的。

In this paper, wo present a method, which recognizes the handwritted Chinese characters by stroke feature matrixes. The stroke width for the inputed characters is to be determined, and the subpattern. of four directions are extracted. Using the centre of gravity for characters pattern, the pattern is divided into several regions. According divided regions, the stroke feature matrix of subpattern regions is maked. The matrix will match with a database for example characters and handwritted Chinese characters are recognized. The simulation experiment shows the method is effectively.