基于三角范数的多模态手指生物特征的分数层融合方法

Finger Multimodal Biometric Verification based on Score-level Fusion

多模态的生物特征融合已成为目前生物识别研究领域的主要发展趋势,从防伪性角度出发,满足普适性和易采集、易接受性的应用需求,提出了基于手指静脉,并结合指纹、指关节纹和指形的分数层融合来实现多模态生物特征的身份认证方案。实验结果表明,本文提出基于Sugeno-Weber三角范数的分数层融合方法,能够增大类内与类间匹配分数分布间的距离,提高了多模态生物特征的身份认证性能。

Multimodal biometrics has become the main trend in the field of current biometric research. From the anti-spoofing point of view,the paper proposes the finger multimodal biometric approach that combines finger vein,fingerprint,finger knuckle print and finger shape to meet the actual demand of universality,acquisition and acceptance convenience. The experimental results show that the proposed score-level fusion approach using triangular norm can enlarge the distance between intra-class and inter-class matching scores and make verification performance improvement over the existing score-level fusion approaches.