一种基于骨架特征和神经网络的手写体字符识别技术

A Kind of Manuscript Character Recognition Technique Based on Framework Character and Neural Network

对英文字符集的识别问题 ,提出了一种依据细化后的字符骨架特征进行分类、并对每一类别的字符各自建立神经网络进行训练和识别的方案。在二值图象理论的基础上 ,通过对字符轮廓特征的分析 ,提出了一种对英文字符集的分类方案 ,同时提出了一种快速有效的骨架链码形成和特征提取算法。

To settle the recognition task of English characters,the authors put forward a scheme as a whole, which advanced sorts the characters based on their figure feature and builts Neural Network to train and recognize for each sort of characters.Based on the theory of Binary Image, After the research and analysis of the characters'figure character,the authors put forword a scheme of sorting the English characters, and put forword a quick and efficient algorithm which named the framework chain code developing and feature getting algorithm.