| 基于HMM/DTW两级结构的签名认证方法 |
| |
| 引用本文： | 于宏斌,吴仲城,申飞.基于HMM/DTW两级结构的签名认证方法[J].模式识别与人工智能,2010,23(1). |
| |
| 作者姓名： | 于宏斌 吴仲城 申飞 |
| |
| 作者单位： | 1. 中国科学院合肥智能机械研究所,仿生感知与控制研究中心,合肥,230031
2. 中国科学院合肥智能机械研究所,仿生感知与控制研究中心,合肥,230031;中国科学院强磁场科学中心,控制组,合肥,230031 |
| |
| 摘 要： | 提出一种基于HMM和DTW在线手写签名认证方法的改进方法.该方法使用签名关键点和关键点的特征值进行签名的状态划分和状态匹配,实现类内签名状态划分的一致性.并利用在线手写签名二维信息的DTW距离作为签名隐马尔科夫模型的状态观测值,构建二级签名隐马尔科夫模型认证框架进行签名认证,得到较好的认证效果.实验结果表明,认证的准确率能达到93%左右.
|
| 关 键 词： | 在线签名认证 动态规整(DTW) 隐马尔科夫模型(HMM) 签名分段 |
Signature Verification Algorithm Based on HMM/DTW Two-Level Architecture |
| |
| YU Hong-Bin,WU Zhong-Cheng,SHEN Fei.Signature Verification Algorithm Based on HMM/DTW Two-Level Architecture[J].Pattern Recognition and Artificial Intelligence,2010,23(1). |
| |
| Authors: | YU Hong-Bin WU Zhong-Cheng SHEN Fei |
| |
| Abstract: | A method for online hand-writing signature verification is proposed based on hidden Markov model (HMM)combined with dynamic time wrapping(DTW).To satisfy the inter-class consistency of signature segmentation and status division,the critical points in signatures are defined and its features are calculated.The difference values measured by DTW using two-dimension signature shape are used as observations of HMM to build the two-level architecture of HMMs for handwriting signature verification.The accuracy rate reaches about 93%. |
| |
| Keywords: | Online Signature Verification Dynamic Time Warping(DTW) Hidden Markov Model(HMM) Signature Segmentation |
| 本文献已被 万方数据 等数据库收录！ |
|
