Sergey Tulyakov

Fingerprint Indexing

Fingerprint indexing is a very important topic in fingerprint matching. Since fingerprints are highly reliable biometrics and widely used, it would be very desirable to use them in identification applications. Indexing will allow fast matching of the query fingerprint against a large set of enrolled fingerprints. So far this is an unsolved problem, and matching query fingerprint against a database of N enrolled fingerprints requires N separate matchings of two fingerprints. Existing approaches can only reduce the matching set to some predefined fraction of the whole fingerprint set.

In our research we propose to construct a single index tree for all enrolled fingerprints, and the matching of the query fingerprint against a database is based on searching the index tree. The tree nodes are based on the minutia subsets.

We expect the matching time to be equivalent to the matching time of two fingerprints; thus instead of O(N) we expect the matching time to be O(1). The project is currently in implementation stage.