User identification using wavelet features of hand geometry graph

Biometrics´ based secure identification and personal verification solutions are essential for today´s information systems. Hand geometry is one of biometric trait that has been widely employed in biometric identification systems. In this paper an innovative peg-free hand geometry based user identification system using the wavelet energy features of a graph representation of the hand is proposed. The user hand is represented as weighted undirected complete connected graph. The graph characteristics are represented as features vector comprising novel zone-wise wavelet energy features of the weighted adjacency matrix of the graph. User identification is performed using multiclass support vector machine (SVM). The proposed method is evaluated on a database of 144 users, with 10 right hand images of each user from GPDS150 hand database. The experimental results demonstrate a correct identification rate of 97.92% using the SVM classifier.