| Fingerprint recognition is one of the most reliable personal identification methods. It plays an important role in criminal justice applications, as well as in non criminal applications such as accessing control or social service databases, where an individual must be prevented from using multiple aliases. We have implemented an indexing-based fingerprint recognition system which shares some ideas with the geometric hashing technique. The system has three main phases: a preprocessing phase, which is performed off line and whose goal is to store information about fingerprints into a database; a recognition phase where an unknown (query) fingerprint is presented to the system and a list of possible matches (hypotheses) is produced by searching the database; and a verification phase where each hypothesis is verified in order to determine if two fingerprints are from the same finger. The system has been tested on a set of 300 fingerprints, illustrating good performance and fast response time. |
