-
摘要: 本文提出一种新的基于指横纹的生物特征认证方法. 指横纹图像具有抗噪性强、纹路简单、可分性强的优点, 并易于与其它手部特征如手形、掌纹等特征形成融合认证系统. 本文采用基于 Gabor 滤波的方法提取指横纹特征点, 并基于互相关点匹配与决策级分数融合完成在线认证系统. 评估系统建立在包含 98 个人、1971 幅图像的数据库上, 平均错误率仅为 0.57\%, 验证了指横纹作为一种生物特征的可靠性与可行性, 同时也证实了认证方法的有效性.Abstract: A novel biometric defined as ``knuckleprint'' is presented in this paper. Knuckleprint preprocessing, Gabor filter based feature point extraction, normalized cross-correlation based feature matching and decision fusion scheme are integrated to implement a real-time verification system. The system is evaluated based on the database that contains 1971 image samples from 98 individuals. The half total error rate (HTER) reaches 0.57\% in the experiments, which clarifies that knuckleprint is reliable and feasible as a biometric, and demonstrates the effectiveness of the proposed method.
-
Key words:
- Knuckleprint /
- Gabor filter /
- normalized cross-correlation /
- score level decision fusion
-
计量
- 文章访问数: 3124
- HTML全文浏览量: 40
- PDF下载量: 1769
- 被引次数: 0
下载:
