手写汉字识别弹性网格Gabor特征提取方法的研究  

Study on Gabor Filter-based Handwritten Chinese Character Feature Extraction

作　　者：金连文[1] 覃剑钊[2] 

机构地区：[1]华南理工大学电子与信息学院,广东广州510641 [2]华南理工大学自动化科学与工程学院,广东广州510641

出　　处：《计算机应用研究》2004年第12期163-165,共3页Application Research of Computers

基　　金：国家自然科学基金资助项目 (60 2 750 0 5) ;广东省自然科学基金资助项目 (0 1 1 61 1 ) ;Motorola研究基金资助项目(B3 83 1 0 )

摘　　要：结合弹性网格技术来对手写体汉字提取Gabor特征,并与基于非线性归一化的Gabor特征及传统的方向特征进行了比较研究,实验结果表明:提出的弹性网格Gabor特征优于非线性归一化Gabor特征,识别性能最稳定;弹性网格Gabor特征要明显优于传统的方向线数特征;不同的非线性归一化方法及归一化的图像尺寸对特征提取的性能有较大影响。New elastic meshing based on Gabor feature extraction for handwritten Chinese character recognition is proposed in this paper. And the effect of different nonlinear normalization methods and different meshing techniques for Gabor feature extraction approach is studied. From experiments, it is found that:The proposed elastic meshing based on Gabor feature is better than conventional directional feature.Elastic meshing Gabor feature is much better and more robust than nonlinear normalization based on Gabor features.Normalization size plays important role in Gabor feature extraction.

关 键 词：GABOR滤波器 手写体汉字特征提取 非线性归一化 弹性网格 

分 类 号：TP391.44[自动化与计算机技术—计算机应用技术]