摘要
介绍了光学字符识别的几种方法以及神经网络的特点 ,神经网络技术能够解决传统OCR方法所不能解决的问题 ,同时指出了手写数字识别存在的困难 ,论证了利用神经网络技术解决这种困难的可能性。本文实现了通过一个含有 1个隐藏层的BP网络来识别手写数字 ,并取得了良好效果 ,论证了这种技术用于手写数字识别的可行性。
Some methods of OCR and feature of Neural Networks(NN) are introduced, and NN technique can solve problems which can't be solved by traditional OCR methods. At the same time, the exist difficult in handwritten digit recognition are pointed, and the possibility of solving the problems using NN technique is proved. NN approach for handwritten digit recognition being proposed in this paper and a one hidden layer BP networks being constructed, more effect are received and the feasibility is proved.
出处
《华东地质学院学报》
2003年第4期383-386,共4页
Journal of East China Geological Institute
关键词
神经网络
模式识别
手写数字
artificial neural network(ANN)
pattern recognition
handwritten digit
