Contributions to hardware implementation of biometric recognition algorithms based on fingerprints (contribuciones a la implementación hardware de algoritmos para reconocimiento biométrico basados en huellas dactilares)

• Autores: Maria Rosario Arjona Lopez
• Directores de la Tesis: Iluminada Baturone Castillo (dir. tes.)
• Lectura: En la Universidad de Sevilla ( España ) en 2014
• Idioma: español
• Tribunal Calificador de la Tesis: Ángel Barriga Barros (presid.), Mariano López García (secret.), Francisco Javier Hernando Pericás (voc.), Carmen Sánchez Avila (voc.), Miguel Ángel Ferrer Ballester (voc.)
• Materias:
  • Física
    • Electrónica
      • Circuitos integrados
  • Ciencias tecnológicas
    • Tecnología electrónica
      • Diseño de circuitos
• Texto completo no disponible (Saber más ...)
• Resumen

  • This Dissertation proposes a new distinctive and compact fingerprint feature, named QFingerMap, which offers competitive performance and requires very low computing cost to be extracted from a fingerprint image and to be matched with other QFingerMaps. A dedicated hardware architecture with an efficient design of all the constituent blocks is presented to carry out the feature extraction. QFingerMap matching is as simple as the comparison of bit strings. Processing time measured for feature extraction in FPGAs from Xilinx is less than 1 millisecond for standard fingerprint image sizes while matching time is negligible (a few nanoseconds). Concerning memory resources, a QFingerMap needs a bit more than 100 bytes. This Dissertation describes how the proposed fingerprint feature can be applied to intelligent fingerprint acquisition systems (which can interact with the user to obtain high quality fingerprint images), fingerprint indexing systems, multi-biometric and two-factor systems (which combine biometric data and passwords), and fingerprint data protection systems. Hardware realizations for all these systems are described, thus showing they can be implemented in one microelectronic device such as a FPGA or ASIC, with constrained resources in terms of computing and memory. Hence, Authentication-on-card realizations are provided, even with template protection. In particular, realizations in FPGAs from Xilinx are described. The design process has followed a model-based methodology supported by CAD tools from Matlab-Simulink and ISE from Xilinx. This has allowed a complete verification of the systems, from high level (using several fingerprint databases) to physical implementation (using FPGA-in-the-Loop). Contributions of this Dissertation can originate the development of small, portable, cheap and/or secure consumer electronic devices for real-time fingerprint recognition.