一种基于特征级融合的多模态生物特征识别方法

生物特征识别是信息技术领域的研究热点,其中多模态生物识别技术凭借更好的适用性、更高的安全性及更优的性能成为发展趋势。提出了一种融合虹膜特征和掌纹特征的多模态生物特征识别方法,该方法分别提取虹膜及掌纹特征,融合时不同于传统的匹配级融合,而是从特征级融合入手,采用并行特征融合策略,将两特征向量以复向量的形式进行融合,构成复向量空间,并利用酉距离进行匹配决策。实验结果表明此方法比单模生物特征方法在识别性能上有了明显改善,同时与传统匹配级融合策略相比,更有优势,识别效果更好。

A Multimodal Biometric Authentication Method Based on Feature level Fusion

Multimodal biometrics is becoming a development tendency due to its better feasibility,security and performance.A multimodal biometric authentication method fusing iris and palmprint is proposed,which extracted both features respectively and combined them at the feature level.When fusing,a parallel fusion strategy was adopted and unitary distance was applied for match and decision.Experiment results in multimodal database represent that the proposed approach can achieve better recognition performance compared to the single modality biometrics as well as the traditional method based on the matching level fusion.