基于因子分析的文本独立笔迹鉴定分类器

Text-independent classifier for handwriting verification based on factor analysis

针对检材与样本笔迹字符内容较少情况下的笔迹鉴定问题,提出一种基于因子分析的文本独立笔迹鉴定新方法.该方法将影响书写笔迹特征距离的因素划分为书写风格的差异和字符形状结构的差异两类因子,然后通过两因子方差分析,分离出特征距离中的字符因子,通过文本依存方法获得文本独立笔迹鉴别分类器.实验证明,该方法得到一种高效的脱机中文笔迹鉴定分类器,在近似实际情况的笔迹鉴别实验中得到良好的鉴别准确率.

In this paper,a text-independent classifier for handwriting identification based on factor analysis is proposed to address the problem when there are not enough characters in the training and testing handwriting documents.The elements affecting matching distances of handwriting features are decomposed into two classes,character factor and writing factor.Then two-way variance analysis mode is built to train these two factors and separate the character factor from the matching distances.Finally the text-independent classifier is obtained.This method works well in the experiments which are very close to the practical application,and it provides a feasible way for the application of computer handwriting verification.