在线单机采集的人手多生物特征识别

单一生物特征识别方法在实际应用时容易受到限制,系统的识别率低、稳定性差.针对上述问题,提出了一种基于在线单机的手形和掌纹相结合的多生物特征识别方法.对于手形识别,提取手指的相对长度构成特征矢量,采用k近邻分类器和支持向量机分类器相结合实现个人身份的识别,然后利用二维Gabor提取掌纹感兴趣区域(ROI)的纹理方向信息作为掌纹特征,对手形分类结果加以认证.在混合图库上进行试验,二者相结合的识别方法的识别率达到98.65%.实验结果表明,采用手形和掌纹双模态特征识别,可以有效提高系统的安全性和稳定性.

Multi-biological Features Recognition of Hand based on Online Integrated Acquisition

The identification method of single biometric is likely to be restricted in practical application. In addition, the system has a low recognition rate and a bad stability. Aiming at this problem, a multi-biometric identification method was proposed which based on the online integrate for the combination of hand shape and palmprint. For the hand shape identification, the feature vector is constituted by the relative length of fingers, using the combination of k-nearest neighbour classifier and support vector classifier to achieve personal identify, then, extracting the texture direction information of the interested region of palmprint （ROI） as the feature of palmprint to certify the results of hand shape classification. Testing in mixed gallery, the recognition rate can achieve 98. 65% by the combining of the both. The experimental results show that the identification method of combining hand shape and palmprint has an obvious advantage. The system＇s security and stability can be effectively improved.