欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
针对当前生物特征识别系统中没有对模板进行定时更新的问题,提出了一种生物特征识别系统中的模板更新方法。首先描述了自升级和共升级算法;然后利用基于路径聚类的方法完成自更新和共更新性能的理论分析;最后提出了路径聚类融合非监督模板更新算法。在大型DIEE多模式数据集平台上的实验结果表明,该方法能够对生物识别系统中的模板进行有效更新,此外,通过仿真模型表明,共更新的性能优于自更新的性能。
英文摘要：
To solve the problem of template update in biometric recognition system at regular time,this paper presented a template update method in biometric recognition system. Firstly, it described the self-update and co-update algorithms. Then, it made a theoretical analysis of self and co-update performance by exploiting the path-based clustering approach. Finally, it proposed the update method based on path-based clustering and unsupervised template. The experiments on the large DIEE multi- modal data set show that the proposed method can update the template in biometric recognition system effectively. Moreover,the results show that the performance of co-update is superior to that of self-update by a simulative model.
同期刊论文项目
同项目期刊论文
期刊信息
位置：