基于多级分类器和神经网络集成的手写体汉字识别

Handwritten Chinese character recognition based on multi-level classifier and neural network ensemble

为了提高系统的泛化能力,在分析了当前汉字识别最新发展技术的基础上,提出了一种三级识别策略的汉字识别系统模型。第一级,使用传统的外围特征法将待选字进行粗分;第二级,使用笔划密度特征法进行细分;第三级,使用一种基于球领域模型的神经网络集成算法对结果进行最后的确认。模拟算法证明,它可以更进一步地提高系统的泛化能力。

To improve the generalization ability of learning systems,after analyzing the currently up-to-date techniques for Chinese character recognition,a handwritten character recognition system model which has 3 levels is proposed.In the first level,the traditional periphery characteristic is used to class approximately.In the second level,stroke density characteristic is used to class accurately and in the third level,neural network ensemble tools based on sphere neighborhood model is used to give the last output.The simulation algorithm testifies that the generalization ability of learning systems is improved.