电信科学 ›› 2018, Vol. 34 ›› Issue (3): 32-40.doi: 10.11959/j.issn.1000-0801.2018130

• 专题:网络空间安全 • 上一篇    下一篇

一种基于指纹和声纹决策级融合识别方法

章坚武,杨佳佳,吴震东   

  1. 杭州电子科技大学,浙江 杭州 310018
  • 修回日期:2018-03-06 出版日期:2018-03-01 发布日期:2018-04-02
  • 作者简介:章坚武(1961-),男,博士,杭州电子科技大学通信工程学院教授、博士生导师,主要研究方向为移动通信系统、多媒体通信技术、网络安全等。|杨佳佳(1994-),女,杭州电子科技大学通信工程学院硕士生,主要研究方向为生物特征识别、通信与网络安全等。|吴震东(1976-),男,博士,杭州电子科技大学网络空间安全学院副教授,主要研究方向为信息安全、网络安全等。
  • 基金资助:
    国家自然科学基金资助项目(61772162);国家重点研发计划经费资助项目(2016YFB0800201);浙江省自然科学基金资助项目(LY16F020016)

Method of fingerprint and voiceprint based decision level fusion recognition

Jianwu ZHANG,Jiajia YANG,Zhendong WU   

  1. Hangzhou Dianzi University,Hangzhou 310018,China
  • Revised:2018-03-06 Online:2018-03-01 Published:2018-04-02
  • Supported by:
    The National Natural Science Foundation of China(61772162);National Key Research and Development Program of China(2016YFB0800201);Zhejiang Natural Science Foundation of China(LY16F020016)

摘要:

结合指纹和声纹在识别环境中的优良互补性以及两种特征相关性不强的特点,提出一种基于指纹和声纹决策级融合识别方法。在对相应的指纹和声纹分别提取特征和识别之后,使用提出的二次比较选择模型(second comparison of choosing model,SCCM)在决策层上利用分数层数据信息进行辅助分类,根据类别判断是否启用声纹识别,使得双模态融合系统平均数据处理量降低了32.8%。基于FVC2002DB1指纹库和自采集声纹库进行了实验验证。实验结果显示,多模态融合可以有效解决单模态的识别限制,识别率相对单模态提升了将近5.08%和4.60%,且稳定性较高。

关键词: 多生物特征系统, 特征选择, 二次比较选择模型, 决策层

Abstract:

Because of strong complementarity in the recognition environment and the low correlation characteristics of the two features,a fingerprint and voiceprint based decision level fusion recognition method was proposed.After extracting the features and recognition of corresponding fingerprint and voiceprint,the second comparison of choosing model (SCCM) was used to classify at the decision-making level and judge whether to enable voiceprint recognition according to the category,reduced the average data processing volume of the dual-mode fushion system by 32.8%.Experiments applied to FVC2002DB1 fingerprint library and self-collecting voiceprint library.The experimental results show that the multi-biometrics fusion effectively solves the single-modal recognition limitation and the recognition rate improves nearly 5.08% and 4.60% as well as get a more stable system.

Key words: multi-biometrics system, feature selection, second comparison of choosing model, decision-making level

中图分类号: 

No Suggested Reading articles found!