周边方向贡献度在脱机手写女书特征提取中的应用

Application of Peripheral Direction Contribution in Feature Extraction for Handwritten Nü Shu Characters

目前国内OCR及其特征提取技术主要应用于汉字的识别,对于少数民族文字,特别是女书,没有提出一种有效的特征提取算法.在分析全局笔画方向密度特征(G-DCD)和局部笔画方向密度特征(L-DCD)的基础上,将周边方向贡献度特征提取算法应用到女书文字的特征提取.实验结果表明:该方法不仅克服了手写体女书字符不规范而造成的误识率高的缺点,并且识别率较前两种方法有明显的提高.

Domestic OCR and feature extraction techniques are mainly used in Chinese characters at present.There has not been a valid feature extraction algorithm for minority languages,in particular Nü Shu.Based on analysis of G-DCD and L-DCD,we introduced a peripheral direction contribution feature extraction algorithm to extracting feature for Nü Shu.Experimental results demonstrated that this method not only has overcome the shortcoming with the high probability of misrecognition resulted from the non-standard handwritten for Nü Shu,but also improved significantly the recognition rate as compared to the first two methods.