有限集手写汉字识别中的笔画分区矩描述特征

Small Set Handwritten Chinese Character Recognition Using Stroke Partition Moment Features

提出了一种笔画分区矩特征的提取方法。根据汉字笔画分布特点，利用小波变换将汉字分解为4个方向笔画分量，用分区矩分别描述4个笔画子图像，并采用K-L变换对特征进行降维处理。采用该特征对有限集手写体汉字进行识别，初步实验结果表明该方法十分有效。

This paper proposes a feature extraction method based on partition moments and stroke decomposition. According to stroke distributing characteristic of Chinese character, a character image is decomposed into four directional sub images using the wavelet transform. Then, partition moment features are extracted from the four sub-images respectively. After dimensions reduction of original feature using Karhunen-Loeve transform, the feature is used for small set handwritten Chinese character recognition. The result of experiment shows high recognition rate which indicates that the method is very effective.