Biometrics & Biostatistics

Biography

Biography: Mohammad Imran

Abstract

There is a global concern to implement accurate person verification in various facets of social and professional life. This includes banking, travel, medical and secure access to social security services. While biometrics has been deployed with various choices such as face, fingerprint, iris, etc., the importance of higher level of security has influenced two main things. One is of finding newer, more universal biometric traits and the other one is multimodal options. Most of the biometric systems employed in the real-world applications are unimodal. They rely on the evidence of a single source of information for authentication which is easier to install and computationally less hectic. The unimodal systems have to contend with a variety of problems. This, in turn, increases False Acceptance Rate (FAR) and False Reject Rate (FRR). A good system needs very low FAR and very low FRR. This can be achieved by the multimodal system. The multimodal system is a sub-set of multi-biometric system which establishes identity based on the evidence of multiple biometric traits. Thus, in this presentation, we address critical issues in designing a multimodal biometric system, i.e., choices of biometric modalities, feature extraction algorithms and fusion strategies. Complementary and supplementary information acquired by feature extraction algorithms are addressed in our work for their contribution towards the improvement of recognition rate. A fundamental issue in designing a multimodal system lies in fusing the information from sensed data. The fusion methodologies at four different levels viz., sensor, feature, score and decision level have been evaluated for the performance with appropriate fusion rules. Fusion methodologies have been exploited for addressing different combinations of multimodal systems.