基于DTW的在线手写签名验证算法

Online Handwriting Signature Verification Algorithm Based on DTW

提出了一种基于DTW匹配的以签名能量为特征的在线手写签名验证算法.在线签名验证包括数据采集、预处理、特征提取、匹配判决等部分.侧重签名能量特征提取方法的研究.基于D aubech ies小波的方法对签名波形进行分解,提取签名波形在跳变点的能量,抽取15个具有较大能量的跳变点.经DTW比较后,提出了一种新的分类算法.实验表明,对于随机伪造签名,当误拒率为0%时,误纳率小于9%.

The paper proposes an online handwriting signature verification algorithm with signature energy as feature based on dynamic time warping （DTW）. Online signature verification includes data acquisir tion, preprocessing, features extraction, and matching and decision-making process. The study of signature energy feature extraction method is emphasized. The energy of sharp trajectory change point in the signature wave is extracted, by means of Daub echies wavelet decomposition of signature. Then, 15 points with most dominant energies are chosen. Finally, after DTW matching, a new algorithm of classification is put forward. Experiment results show that false acceptance rate is 9 percent while false rejection rate is 0 percent for random forgery.