Minería de datos en computación de altas prestaciones para identificación en base a huellas dactilares Peralta, Daniel Herrera Triguero, Francisco Benítez Sánchez, José Manuel Universidad de Granada. Departamento de Ciencias de la Computación e Inteligencia Artificial Minería de datos Identificación Dactiloscopia Computación de altas prestaciones Proceso electrónico de datose This thesis starts by presenting a deep study of the scientific literature on minutiae-based local matching matching techniques, establishing a taxonomy of the available local structures and consolidation methods, and highlighting the main advantages and drawbacks of each of them. Then, we will present a minutiae filtering algorithm that removes spurious or misleading minutiae to improve both the identification time and the accuracy of the recognition process. After that, we will describe two frameworks for massively parallel fingerprint identification, which are able to execute diffierent matching algorithms adapting to the underlying hardware for maximum performance and full scalability. We will also develop a framework to combine the information of two fingerprints and the capabilities of two diferent matching algorithms to address both problems that hinder identification in large databases: the high identification time and the loss of accuracy. Finally, we describe a new classification strategy to reduce the penetration rate of the identification. Finally After this introduction section, Section 2 describes in detail the background of the main areas addressed in this thesis: fingerprint feature extraction (Section 2.1), fingerprint identification (Section 2.2), high performance computing (Section 2.3), database penetration reduction and fingerprint classification (Section 2.4) and information fusion for fingerprint identification (Section 2.5). After that, Section 3 presents the justification of this memory, describing the open problems addressed throughout this thesis. The objectives pursued to address these problems are detailed in Section 4, along with the methodology followed along the thesis in Section 5. Section 6 summarizes the works that compose this memory, while Section 7 presents the results obtained in them, performing an analysis in relation with the tackled objectives and how they have been reached. Section 8 presents the conclusions after the work carried out for this thesis. Finally, in Section 9 we point out several future lines of work that have been derived from the results achieved. 2017-01-30T11:10:02Z 2017-01-30T11:10:02Z 2016 2016-09-26 doctoral thesis Peralta Cámara, D. Minería de datos en computación de altas prestaciones para identificación en base a huellas dactilares. Granada: Universidad de Granada, 2016. [http://hdl.handle.net/10481/44550] 9788491630272 http://hdl.handle.net/10481/44550 eng http://creativecommons.org/licenses/by-nc-nd/3.0/ open access Creative Commons Attribution-NonCommercial-NoDerivs 3.0 License Universidad de Granada