摘要
身份认证是保证信息与网络安全的一种重要手段,手形认证是身份认证的重要方法之一。传统手形识别方法大致分成特征矢量匹配和点模式匹配两种:前者通过计算手形的长度和宽度等特征矢量来对不同手形进行匹配认证,该方法计算量小,但是误识率偏高;后者通过将手形轮廓图象表示为一系列特征点集,然后对两个手形的特征点集进行匹配认证,误识率较小,但计算量和拒识率相对较大。以上原因导致了两种算法都不能被广泛应用。该文提出了一种基于模板的点匹配算法,可以较好地解决点模式匹配计算量过大的问题,同时也能够提高认证识别率。在认证过程中还采用了方向角及膨胀收缩修正等方法,使得模板的匹配速度和拒识率得到有效的改善,从而大大增强了认证过程的鲁棒性。
Reliability in the personal authentication is key to the security in the networked society.Many Biometrics features have been suggested for the security in access control.Hand Shape Verification(HSV)is one of major biometric methods in identity verification.The traditional HSV approaches can be classified into two major kinds:one is Character-istic Vectors Matching(CVM)method and the other is Point Pattern Matching(PPM)method.CVM takes the verification procedure by calculating the length and width of the hand.It features less computation burden but shows high Error-Accept-Rate(EAR).PPM turns the hand shape image into a set of characteristic points and makes the verification ac-cording to the point set matching.As compared to CVM,PPM features low EAR but high Error-Reject-Rate(ERR)and computation complexity.Therefore,CVM and PPM cannot be applied widely.In this paper,we propose an novel Template-based Point Matching Algorithm(TPMA)to overcome the computation complexity disadvantage of PPM and get better per-formance in HSV.During the process of verification,we adopt directional angle and expanding-contracting modification operation to effectively speed up the matching and lower the ERR and greatly enhance the robustness of our verifica-tion algorithm.
出处
《计算机工程与应用》
CSCD
北大核心
2005年第6期85-88,224,共5页
Computer Engineering and Applications
关键词
手形认证
模板匹配
点模式匹配
特征矢量法
最大相似度
Hand Shape Verification(HSV),template matching,Point Pattern Matching(PPM),Characteristic Vectors Method(CVM),maximum similarity