Fingerprint recognition systems based on smartphones

Fingerprint-based biometric systems are used in many application areas for different purposes, such as developing high-reliability platforms or detecting criminals in forensic sciences, and perform very critical tasks. In this thesis, two different system proposals are presented, called MAFIS and MALFIS, which serves the application areas where fingerprint-based biometric systems are used. MAFIS is a system that can collect fingerprint from the individual without causing data loss, noise, deformation and without being affected by the disadvantages of touch-based acquisition methods, and identify/verify by using this trace. MALFIS, on the other hand, is a system that can collect the latent fingerprint left by the suspect at the crime scene, using non-destructive and contactless methods, without the need for chemical and/or physical intervention, and make identification using this trace. MAFIS and MALFIS have been specially developed to work on smartphones, and have data acquisition mechanisms for which two different apparatuses are used specially designed for fingerprint detection, development and collection. In order to determine whether the fingerprints collected from both systems using these mechanisms are utilisable for identification/verification, these fingerprints are subjected to pre-processing and make ready for the matching. MAFIS and MALFIS are end-to-end solutions that solve biometric systems' problems such as not being portable, high-cost design and technology requirement, inclusion of complex and specialized application steps. In order to analyze the performances of proposed systems, the individuals' fingerprints collected using traditional methods and using the proposed systems have been compared using matchers accepted in the literature (VeriFinger-SDK and SourceAFIS), the matching successes have been evaluated, and the accuracy rates obtained for MAFIS and MALFIS have been measured as 98% and 96%, respectively. These findings demonstrate that proposed systems are acceptable in terms of identification and verification accuracy. It is considered that these systems might contribute to the development, acceleration and reduction of costs of the processes used in identification/verification, and most significantly, and reduce the need for experts.