Differential Synthetic Biometric Sample and Template Generation

Creation of large databases of synthetic biometric images to emulate and model system performance for identification is highly dependent on processing and encoding techniques resident in large-scale systems. To this end, IBG will conduct research to support development of software that generates synthetic biometric data. The software and its associated algorithms will incorporate image and feature data most relevant to the primary biometric identification modalities (fingerprint, face, iris) and the most commercially relevant techniques. This approach will maximize the variability in both atomic and clustered biometric templates in order to model performance at the extremities contemplated by biometric matchers. The proposed research will build on the state of the art of forensic biometric data analysis, physiological and behavioral analysis, and assessment of imaging and illumination technologies to develop a vendor-agnostic data generation capability. By balancing the conceptual limits of differentiated human physiology with a practical understanding of the manner in which biometric databases and indexes are created and exploited, the research will provide substantial short-term commercial benefits. The commercial applications for this software include licensing to biometric vendors, systems integrators, and algorithms developers involved in large-scale biometric system design, development, and implementation.