基于小波变换和ART网络的手写数字识别  

Handwritten Digit Recognition Based on Wavelet Transform and ART Neural Networks

作　　者：张捷[1] 寇雪芹[2] 封俊红[1] 

机构地区：[1]西安建筑科技大学管理学院,陕西西安710055 [2]西安建筑科技大学机电学院,陕西西安710055

出　　处：《计算机测量与控制》2004年第11期1093-1095,共3页Computer Measurement &Control

基　　金：陕西省教育厅专项基金资助项目(01JK201)

摘　　要：由于小波变换能有效地提取字符的结构特征,自适应共振(ART)网络有很好的学习能力。将二者结合起来,用小波变换抽取特征、用自适应共振ART网络作模式分类器来识别手写数字。实验证明该方法有很高的识别率,能够有效地进行手写数字的分类,可以满足实际应用。Because of that wavelet transform can effectively extract the features of character construction and adaptive resonance theory (ART) neural networks has a nice learning ability, the two aspects are combined to recognize handwritten digit using wavelet transform to extract features and adaptive resonance theory (ART) neural networks for classification. The result of experiment shows high recognition rate, which indicates that the methods can effectively classify hand written digit and be put into practical use.

关 键 词：手写数字识别 小波变换 自适应 字符 识别率 网络 模式分类器 抽取 共振 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TP393.09[自动化与计算机技术—计算机科学与技术]