节点文献
手写体汉字识别中小波分形分解特征的研究
Feature extraction with wavelet and fractal for handwritten Chinese character recognition
【摘要】 研究了手写体汉字识别中的一种新的特征提取方法——小波分形分解特征。对手写汉字分别采用小波和分形的方法提取其结构特征和统计特征,并将提取的结构特征和统计特征组合后作为识别器的输入进行识别。实验结果表明,对训练样本可以达到98.71%识别率,对测试样本可以达到91.37%识别率。
【Abstract】 A new feature extraction approach for computer recognition of handwritten Chinese character is studied. In this method, wavelet and fractal are used to pick up the structure feature and statistical feature, and these features are combined and used for recognizer as input data. The experiment shows satisfied result. Identify rate can reach 98.71% for training stylebook and 91.37% for testing sty- lebook.
【关键词】 汉字识别;
特征提取;
小波;
分形;
【Key words】 Chinese characters recognized; feature extraction; wavelet; fractal;
【Key words】 Chinese characters recognized; feature extraction; wavelet; fractal;
【基金】 山东省自然科学基金项目(Y2001G03)
- 【文献出处】 计算机工程与设计 ,Computer Engineering and Design , 编辑部邮箱 ,2005年05期
- 【分类号】TP391.41
- 【被引频次】11
- 【下载频次】170