A Score Level Fusion Approach for a Multimodal Biometric Identification System Using Ear and Palmprint

Abstract

Today, security is one of the most important challenge faced everyday’s human life. One of the applicable security procedures is human identification using biometrics. Biometric is the Knowledge of using behavioral or physiological characteristics to identify individuals. The limitations of unimodal systems such as sensor noise, restrictions, degrees of freedom and piracy attacks has led researchers to investigate the combination of these systems to have a multimodal system. These systems usually consist of four levels of data acquisition, normalization, feature extraction and classification and finally the decision. A multimodal biometric system is obtained fusing the unimodal systems in each of these levels. In this paper, using a new algorithm in the hierarchical score level fusion, a multimodal biometric system is proposed based on ear and palmprint images. Gabor filter is used to extract textural features from the normalized images and coding process is carried out on the results. The scores are obtained using matching from two samples by hamming measure and finally the proposed fusion is applied. To evaluate our proposed algorithm, USTB and PolyU databases for ear and palmprint are used. The obtained results of ERR=0.91% shows that our approach has good accuracy compared with existing algorithms in the literature.