High Confidence Multimodal Biometric System

Phase I consists of a detailed feasibility study examining the potential of multimodal biometric systems to provide more robust accuracy and throughput than single-modal biometric systems. An experimental system capable of utilizing previously acquired, correlated data - match scores, templates, and samples will be designed and built. This system utilizes combinations of two biometric modalities (selected from face, fingerprint and iris) for the purpose of experimentation with multiple data fusion models. Phase I outputs will include data indicative of the degree to which alternative fusion models improve identification and verification performance as well as strategies to extend the methodology to a tri-modal system for Phase II. Phase II extends the Phase I experimental system to encompass development of an end-to-end prototype multimodal system. This hardware and software system enables live acquisition of biometric data from subjects for use in online identification and verification modes. Phase II further builds on Phase I by enabling more than two biometric modalities at a time.