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.
出处
《计算机工程与设计》
CSCD
北大核心
2008年第20期5379-5382,共4页
Computer Engineering and Design
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
handwritten digit recognition
topological features constitution
classification tree
preprocessing
feature extraction
