脱机手写数字识别方法  

Method of off-line handwritten digit recognition

作　　者：杨金伟[1] 段会川[1] 

机构地区：[1]山东师范大学信息科学与工程学院,山东济南250014

出　　处：《计算机工程与设计》2008年第20期5379-5382,共4页Computer Engineering and Design

摘　　要：脱机手写体数字识别有着重大的使用价值,特征提取占据了重要的位置。提出了一种通过拓扑特征构造的特征提取新方法,利于了9种特征对数字进行特征提取,然后利用分类树的方法将数字进行分类。最后,在本科学生手写数字图像样本库上的试验结果表明,提出的特征提取方法不仅具有很快的运算能力,而且较大幅度地提高了识别率。Off-line handwritten digit recognition has great value, in which feature extraction occupies important positions. A new method of the topological features constitution is proposed, and nine kinds of features is used to extract digital features. Then we use classification tree approach to classify digits. Finally, the test is done based on undergraduate students＇ handwritten samples of the digital images, and the test results show that this feature extraction method we proposed not only has the rapid computing capacity, but also more substantially improves the recognition rate.

关 键 词：手写体数字识别 拓扑特征构造 分类树 预处理 特征提取 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]