Multi-Biometric Recognition System Based on a Specific Set of Features

Biometric techniques are replacing traditional passwords and Personal Identification Numbers since they are more dependable and secure. Unimodal biometric systems have various challenges, including non-universality and noisy data. The systems that use multiple modalities of biometrics can help overcome some of these restrictions, but retrieval and matching processes are time-consuming due to the high-dimensional nature of biometric data. This research proposes a serial multi-biometric recognition system using iris and face models based on only 16 features. Iris features are retrieved after splitting the region of interest into (2 × 2) blocks and computing the mean, standard deviation, and energy values for each block. While the moment’s equations are calculated for face feature retrieval. In order to determine a person’s identity, the iris template is matched using the Euclidean distance to existing templates in the database, and matches the face feature only to the near candidate persons from the iris features. For iris and face images, respectively, the Multimedia University (MMU) and Faces94 datasets were used to assess the system’s performance. The outcomes demonstrate how well the proposed system performs in increasing the identification rate.