A Novel Feature Extraction-based Human Identification Approach using 2D Ear Biometric

Biometric is the study of physiological or behavioral characteristics of human being to authenticate his/her identity. In literature, there are a limited number of studies, which have used ear biometric for human authentication or recognition, of them, some are either scale invariant or shape invariant. In addition, time complexity for the entire recognition process is also reported high. To address these issues simultaneously, a novel algorithm has been proposed in this study, which helps in efficient feature extraction. In this extraction, edges from the ear template are searched, and are fitted with appropriate lines. Thereafter, the ear template with fitted line segment is folded in two dimensions (horizontal and vertical). Angles formed in the folded ear template are used to extract different statistical information. To recognize a human, three main features, namely the number of edges, the number of bifurcation points, and the angle information are considered in this study. The experiment is carried out using a total of 1200 ear images obtained from 120 people. Two classifiers, namely minimum distance (MD) and K-Nearest Neighbor (K-NN) are used for recognition. Results reveal that the use of proposed feature extraction technique helps to obtain higher classification accuracy in human identification.