基于信息融合的在线手写签名算法研究

Research of on-line handwriting signature verification algorithm based on information fusion

为了进一步提高认证效果,在演化计算、神经网络和离散F距手写签名认证算法的基础上,提出了基于信息融合的在线手写签名认证算法。该算法将测试签名和参考签名分别通过三种算法进行认证,得出测试签名为真实签名的置信度,然后对三种认证算法的结果进行加权融合,根据最终的融合结果进行签名真假的判定。实验结果表明,信息融合算法的误拒率和误纳率都有显著的减少。

In order to improve this verification effect, the paper proposed the algorithm of on-line handwriting signature verification based on information fusion.The algorithm fused three kinds of algorithms of on-line handwriting signature verification respectively based on evolution computation, neural network and discrete Fréchet distance. Verified the test signature by the three kinds of algorithms to obtain its confidence level of every algorithm. Fused the last confidence level of the signature by the confidence level of every algorithm by calculating their weights, and it could be used to judge whether the test signature was genuine. The experiment results reveal the remarkably decreasing percentage of false rejection and false acceptance.