摘要
使用了一种新的神经网络模型———动态衰减调节径向基函数(RBFDDA),并结合一种新的特征提取方法来进行手写体汉字识别的研究,通过对100种汉字、15000个样本的初步实验,取得了识别率为99%的良好结果,表明将RBFDDA引入到手写体汉字识别的研究是比较成功和可行的。
A novel neural networks model—Dynamic Decay Adjustment Radial Basis Function Networks (RBF DDA) together with a new feature extraction method is introduced to Handwritten Chinese Character Recognition (HCCR).Experiments on 100 categories of handwritten Chinese characters produced the recognition rate of 99%,showing that the proposed approach for HCCR is promising.
出处
《华南理工大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
1997年第9期97-101,共5页
Journal of South China University of Technology(Natural Science Edition)
基金
华南理工大学自然科学基金
关键词
径向基函数神经网络
动态衰减调节
手写体汉字识别
特征提取
Radial Basis Function neural networks
dynamic decay adjustment
handwritten Chinese character recognition
feature extraction