Multimodal biometrics system using feature-level fusion of iris and fingerprint

Biometrics system had been widely implemented in our daily life applications. With continuous improvement in biometrics recognition performance, biometrics security hence becomes an important topic of research as biometric template protection scheme serves as a vital part of a complete biometrics system. Besides, multimodal biometrics system is introduced to improve the recognition performance, system complexity, security and applicability of nowadays biometrics applications. In this paper, we present a new approach of feature-level fusion multimodal biometrics system using indexing-first-one (IFO) hashing and integer value mapping strategy. Indexing-first-one hashing has proven survived from several major privacy attacks such as single-hash attack (SHA), attack via record multiplicity (ARM) etc. On top of that, a weighted feature level fusion approach is proposed where multiple biometrics are given different weights based on the individual recognition result which then each biometrics will contributes to the final matching result based on their respective weights. The experiment is conducted and result is validated using a multimodal fingerprint and iris database.