基于改进多数规则的笔迹鉴定信息融合

Handwriting verification information fusion based on advanced majority rule

针对传统笔迹鉴定信息融合算法的不足,提出了改进多数规则信息融合算法。在笔迹鉴定过程中,分别用文本依存的距离变换法、几何矩算法和多通道分解与匹配算法三种特征提取算法进行特征提取,三种算法的提取结果采用基于改进多数规则的笔迹鉴定信息融合算法进行融合。在融合的同时,单独对每种笔迹鉴定方法在相同的条件下进行了实验,并且设计相应分类器,估计分类器的错误率。实验结果表明,运用本文的融合算法,在融合后笔迹鉴定准确率有大幅度的提高。

For the insufficiency of traditional information fusion algorithms used in handwriting verification,advanced majority rule information fusion algorithm is put forward.Three feature extracting algorithms which are distance transform algorithm,geometric moment algorithm and multichannel decomposition and match algorithm are used in the process of handwriting verification.The results of three algorithms are fused by the handwriting verification information fusion algorithm based on advanced majority rule.Meanwhile,the experiment of each handwriting verification method is made in the same situation,the appropriate sorter is designed and the error rate of the sorter is estimated.Experimental results indicate that compared with current methods,verification accuracy is greatly raised with this approach.