Uma Nova Abordagem de Casamento de Impressões Digitais

Abstract

Biometric recognition has been increasingly adopted in several applications. The main reason for this choice is based on the fact that biometric information is harder to forge, share or misplace in comparison with other authentication alternatives. Many individual traits can be used in a biometric system, such as fingerprint, face, voice, iris and others. Several researches have explored these and other traits producing safer and more accurate recognition methods, but none of them are completely reliable and there are much to improve in this area. This work presents a new approach for fingerprint matching, which is an important step in the identification process. Many comparison alternatives between fingerprints have been used aiming to improve the reliability of the proposed method. This work evaluates two bio-inspired algorithms for parameter optimization, Genetic Algorithms and Particle Swarm Optimization. To validate the proposed method the Equal Error Rate of six databases have been measured, including databases from previous Fingerprint Verification Competitions and a local database named BioPass-UFPB. Results presents 0.2% Equal Error Rate.