Diseño e Implementación de un Sistema de Seguridad para un Automóvil con Autenticación por Reconocimiento Facial Utilizando Técnicas de Visión Artificial.

Abstract

In Riobamba city to determine an appropriate camera, a facial recognition algorithm, and a proper embedded system, the researcher designed and assembled a car security system with facial recognition authentication a microcontroller with ARM 32-bit architecture free software OPENCV that is used to analyze images and whether to provide the system of artificial vision on a small computer Raspberry Pi 2 has the Raspbian operating system. The system has different parts or stages; it has a module for geo location consisting of an Atmega 328p with a module GSM / GPRS / GPS SIM 908 to handle wireless networks such as GSM and GPS. The system has the ability to maintain without starting the vehicle if authentication does not overcome, it can warn us of anomalies with a missed call; it is able to give us your approximate location for short text messages. The system has an Android application to interact with the user which has the option to open the door lock of the automobile through commands sent via Bluetooth, also it has super user key for emergency cases. The camera USB webcam type got results from 74% in contrast to the type of higher demand, it isn`t the same with the Internet Protocol cameras (IP) and with Serial Interface Connector (CSI) with higher percentages; likewise, the Local Binary Patterns Histogram (LBPH) method, it obtained minor requirements with a 85.57%. The conclusion is that the best camera and the best method of face recognition proved to the USB webcam and LBPH respectively, also considering that the most appropriate embedded system was the aforementioned Raspberry Pi 2. These results recommend applying those used in this case, in order they don`t fall into unnecessary oversizing in the application.