基于组合分类器的自由手写体数字识别方法  

Unconstrained handwritten numerals recognition way based on combined classifier

作　　者：傅德胜[1] 谢忠红[1] 苏坚[1] 

机构地区：[1]南京气象学院计算机科学与技术系

出　　处：《计算机工程与设计》2004年第10期1713-1715,共3页Computer Engineering and Design

基　　金：南京气象学院科研基金(Y301)

摘　　要：自由手写体数字识别广泛应用于信息录入和文本识别中。基于组合分类器实现手写数字的识别,克服了单因子识别的局限性,识别中使用距离法和改进的BP神经网络方法,以多种特征向量作为分类器的输入,以举手法则确定识别输出。实验证明,该系统具有较高的识别率和极低的误识率,有令人鼓舞的应用价值。A combined classifier is designed to recognize the handwritten numerals., The traditional least distance recognition way and the improved BP neural network are used in the combined classifier. Many character vectors are input to the combined classifier. The hand-rise arithmetic is used to get the recognition results. The results of the experiment show that the system has high recognition rate and low refusing rate. It has inspiring application value in our life.

关 键 词：分类器 手写体数字识别 手写数字 识别率 录入 误识率 输入 组合 克服 法则 

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