Publication: Implementación de mecanismos biométricos para autenticación de usuarios en aplicaciones multidispositivo
Loading...
Official URL
Full text at PDC
Publication Date
2022
Advisors (or tutors)
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Citations
Exportar
Abstract
La mayoría de aplicaciones modernas cuentan con diferentes formas de identificación para acceder a sus servicios, aunque el uso de contraseñas es la forma más extendida en la actualidad. La autenticación basada en el reconocimiento biométrico puede solventar varios de los problemas que conlleva el uso de contraseñas, y como consecuencia, mejorar la experiencia de los usuarios. Por desgracia, hoy en día el uso de la biometría no es muy común, especialmente en computadores tradicionales (ordenadores de sobremesa o portátiles). Partiendo de dicho contexto, este proyecto centra sus esfuerzos en emplear el reconocimiento facial para la autenticación de individuos. Esta forma de reconocimiento divide su funcionamiento en dos actividades principales: la detección facial y el reconocimiento facial propiamente dicho. Se ha realizado el análisis de ambas fases tanto a nivel teórico como experimental, con el fin de escoger aquellos modelos que presenten mejores resultados e incluirlos en el marco de una aplicación multidispositivo. Hay que destacar el algoritmo de Viola-Jones y la implementación presente en la librería Dlib para llevar a cabo la detección. En cuanto al reconocimiento facial destacan los algoritmos de Fisherfaces, Eigenfaces, LBPH y el modelo actual provisto por Dlib. La aplicación multidispositivo se ha implementado como un servicio web, cuyo desarrollo se ha realizado empleando la herramienta de Django y Python, el lenguaje que más se ha utilizado. Finalmente, se ha demostrado que es posible la implementación del sistema de autenticación en una aplicación web funcional en la que se realiza el reconocimiento facial en tiempo real para confirmar las identidades de los usuarios.
Most modern applications use different forms of identification to access their services, the use of passwords is currently the most widespread form. Authentication based on biometric recognition can improve the user experience by solving some of the problems that come with the use of passwords. Unfortunately, the use of biometrics is not very common nowadays, especially on traditional computers like desktops or laptops. Taking the this into consideration, this project focuses its efforts on applying facial recognition to authenticate individuals. This form of recognition is divided in two main activities: facial detection and facial recognition itself. An analysis of both activities has been carried out both in a theory and an experimental basis in order to determine those models with better results and behavior. The best models are to be included in the framework of a multi-device application. We must highlight the Viola-Jones algorithm and the implementation present in the library known as Dlib to carry out the detection. Regarding facial recognition, the algorithms of Fisherfaces, Eigenfaces, LBPH and the current model provided by Dlib stand out the most. The multi-device application has been implemented as a web service, whose development has been carried out mainly using both Django and Python. Finally, it has been shown that it is possible to implement an authentication system in a functional web application in which facial recognition is performed in real time to confirm the identities of its users.
Most modern applications use different forms of identification to access their services, the use of passwords is currently the most widespread form. Authentication based on biometric recognition can improve the user experience by solving some of the problems that come with the use of passwords. Unfortunately, the use of biometrics is not very common nowadays, especially on traditional computers like desktops or laptops. Taking the this into consideration, this project focuses its efforts on applying facial recognition to authenticate individuals. This form of recognition is divided in two main activities: facial detection and facial recognition itself. An analysis of both activities has been carried out both in a theory and an experimental basis in order to determine those models with better results and behavior. The best models are to be included in the framework of a multi-device application. We must highlight the Viola-Jones algorithm and the implementation present in the library known as Dlib to carry out the detection. Regarding facial recognition, the algorithms of Fisherfaces, Eigenfaces, LBPH and the current model provided by Dlib stand out the most. The multi-device application has been implemented as a web service, whose development has been carried out mainly using both Django and Python. Finally, it has been shown that it is possible to implement an authentication system in a functional web application in which facial recognition is performed in real time to confirm the identities of its users.
Description
Trabajo de Fin de Grado en Ingeniería Informática, Facultad de Informática UCM, Departamento de Ingeniería del Software e Inteligencia Artificial, Curso 2021/2022.