Poster: Secure Computation of Fingerprint Alignment and Matching

Secure computation with biometric data is a challenging topic and attention to this area has grown during recent years. This is because biometric data is highly sensitive and requires protection when used in a variety of applications including biometric verification and identification. Fingerprint images are one of the most accurate type of biometry used in these applications and they are highly sensitive due to their ability to uniquely identify the data owner. For that reason, when fingerprint data is used in applications, it is desired to properly protect such data. Typically the computation involves comparing two fingerprint representations in different settings ranging from comparing two databases belonging to different entities to searching for a fingerprint in an external database or outsourcing fingerprint comparisons for computational reasons. In this work, we put forward three secure protocols for fingerprint alignment and matching based on the most precise or efficient algorithms in the biometric literature. To the best of our knowledge, this is the first time fingerprint alignment based on minutia points is considered in a secure computation framework due to the complexity of the algorithms. After designing the necessary building blocks, we build the overall protocols and realize them in the two-party setting using garbled circuit evaluation and in the multi-party setting using secret sharing techniques. Keywords—Fingerprint alignment, fingerprint matching, secure computation, garbled circuit evaluation, secret sharing.