一种基于交点特征的印刷体数字识别方法

为了进一步提高印刷体的数字识别准确率,提出了一种基于交点特征和径向基函数神经网络的数字识别方法。首先利用交点特征对数字进行特征提取,即提取某一数字的划水平线得到的交点数作为水平特征分量,提取划垂直线得到的交点数作为垂直特征分量,将水平特征向量与垂直特征向量组合成数字的交点特征向量;然后利用径向基函数神经网络学习不同模式类别中的学习样本,学习过程完成后,利用此网络对样本进行识别。实验结果表明,该数字识别方法在印刷体数字识别中正确率可达到100%,处理效果良好。

An Approach to Printed Digital Recognition with Intersection Features

In order to further improve the accuracy rate of printed digital recognition, a new digital recognition approach which combined intersection features and Radial Basis Function (RBF) neural network was proposed. Firstly, the intersection features of numbers were extracted. Namely, the numbers of the intersection points of the number with some dividing lines in horizontal direction were extracted as the horizontal features and the numbers of the intersection points of the number with some dividing lines in vertical direction were extracted as the vertical features. And all of the horizontal features and the vertical features of the number were combined as the intersection features. Secondly, the samples in different modes are trained in the RBF neural network. After training, the samples are recognized in the RBF neural network. The results of experiment show that the recognition rate of printed numbers can achieve 100%. The treatment effect of this approach is good.