A adaptive dorsal hand vein recognition algorithm based on optimized HMM

作者：Jia Xu^*; Cui Jianjiang; Xue Dingyu; Pan Feng

来源：Journal of Computational Information Systems, 2012, 8(1): 313-322.

• Adaptive hidden Markov models
• Authentication methods
• Biometric features
• Hand vein recognition
• HMM
• Identification rates
• Location information
• Number of state
• Parameter optimizing
• Pre-processing
• Radon Transform
• Recognition algorithm
• Recognition methods
• Recognition time
• Vein recognition

As a kind of biometric feature authentication method, vein recognition has more merits than others. In this paper, a new vein recognition algorithm based on adaptive hidden Markov model (HMM), where the parameters of HMM are optimized by using stepper increasing method according to different vein databases, and in the database every vein object can be represented as a HMM. Because the main features of vein image are reflected in direction and location, and the direction and location information of image can be obtained through Radon Transform, Radon Transform is applied to the thinned vein image which was obtained after pre-processing; Different sets of HMM parameters are used in different databases, in the proposed algorithm the parameters, which include the number of states, the number of distinct observable symbols and the number of observations in the sequence, are adjusted by stepper increasing. Experimental results show that the proposed recognition method outperforms two other methods which are based on feature points and image fusion respectively in terms of correct identification rate, and the recognition time 0.850 s is sufficient to meet the requirement of real-time.

• 出版日期2012
• 单位东北大学