基于BP神经网络的手写数字识别算法研究

讨论了一个手写数字识别系统的原理及其实现。特征提取的方法是:计算字体轮廓的曲率特征,并在计算曲率的过程中使用了B样条函数;对曲率进行了大小和平移规整化,这样得到的曲率具有大小和方向的不变性。为了得到更紧凑的特征,采用了小波对其进行降维。采用了BP神经网络作为分类器,实验结果表明,对于字形相似的数字也达到了较高的识别率。还简介了识别系统的模块设计和界面设计。

Research on Handwritten Numeral Recognition Algorithm Based on BP Neural Network

This paper studies the theory of handwritten numeral recognition and its implement.The method of extracting feature is described as follows.Calculate the curvature of the contours of handwritten numerals.B-splines are used in the calculating.The concept of scale and shift normalization is introduced so that the curvature signal can be normalized as scale and rotation invariant.To produce more compact features,wavelet basis decomposition is used to reduce the dimension of the feature.This paper uses BP neural network as a classifier.Experimental result shows a high recognition rate of discrimination of similar handwritten numerals.The module design and software interface of the recognition system are also introduced briefly in this paper.