一种改进的脱机手写汉字四角特征粗分类方法

将粗分类应用于脱机手写汉字识别中,采用这种多层次分类策略,能有效地改善识别的性能,提高识别精度。本文提出了一种利用四角区域结构特征对手写汉字进行粗分类的方法。在对汉字基本笔画进行分析的基础之上,根据手写汉字形变的特点以及识别算法的要求,定义一组新的笔画单元,并将这些笔画单元与汉字特定区域内的结构进行比对,得到一组4位结构特征编码,以此作为脱机手写汉字粗分类的依据。对GB2312一级字库中的部分手写汉字进行采样和识别实验,结果证明改进的四角结构特征用于粗分类的有效性。

An Improved Coarse Classification Method for Off-line Handwritten Chinese Character Based on Features of Four Corners

Use coarse classification in off-line handwritten Chinese character recognition can effectively improve the performance and accuracy of recognition. In this paper present a method that use four corner feature for pre-classification of handwritten Chinese characters. In the basis of analyze the basic strokes of the Chinese characters, according to the characteristics of handwritten Chinese characters deformation as well as the requirements of the recognition algorithm, define a new group of the stroke unit and match them with the structure of Chinese characters in a specific area, to get a the group of four structural characteristics encoded as a basis for offline handwritten Chinese characters coarse classification. Experimental results on 500 Chinese characters handwritten samples show that our method can achieve a satisfied result in coarse classification for off-line handwritten Chinese characters.