Robust multimodal biometric identification system using Finger-Knuckle-Print features

Automatic personal identification is becoming an increasingly important requirement in a variety of applications like access control, surveillance systems and physical buildings. In recent years, biometric systems are used in these fields, offering greater convenience and several advantages over traditional security. In this paper, we propose an efficient online personal identification system based on Finger-Knuckle-Print (FKP) images. In this study, the finger image is characterized by the two dimensional Bloc based Discrete Cosine Transform (2D-BDCT) coefficients. Then, we use the Hidden Markov Model (HMM) for modeling the observation vector. In addition, the same finger is transformed by the two dimensional Discrete Fourier Transform (2D-DFT). The response of this transformation is directly used to create another observation vector. Subsequently, the two subsystems are integrated in order to construct an efficient multimodal identification system based on matching score level fusion. The results on a medium-size database, 150 users, show good identification performance based on individual modalities as well as after fusing multiple finger types.