基于net的核判别分析的手写体识别

Handwriting Recognition Based on Net Nuclear Discriminant Analysis

提出了一种新的基于核判别分析的手写汉字识别方法。核判别是对线性判别式分析的非线性判别分布的扩展。阐述了核判别分析法的基本原理,建立了核判别分析手写体识别模型,研究分析了核判别分析手写体识别模型的缺陷并提出了优化策略。在此基础上,采用C#与核判别分析相结合的算法,更好地展示了核判别算法的算法优势,采用高级语言提高了网络的学习训练速度和识别效果。

A novel handwriting recognition method based on discriminant analysis using Kernel function is introduced. KDA is an extension of LDA to non-linear distributions. First,the principle of KDA is displayed,then the model on KDA for handwriting discriminant analysis is built. Secondly, the algorithm and the advantages and deficients of the model is detailed. Finally, the algorithm is realized by using C # language, which better diplays its advantages. The novel recognition method increases the training online and the recognition speed.