Sistema de reconocimiento facial utilizando el análisis de componentes principales com una red neuronal Backpropagation desarrollada en C# y Matlab

Abstract

This thesis presents the development of a facial identification system based on three fundamentals such as: artificial vision that is a multidisciplinary area, which, through the processing and interpretation of information received by the web camera to reproduce artificially the sense of sight, biometrics and specifically facial biometrics that is a technology that can identify an individual by capturing a facial image, making comparisons of certain characteristics of the face with that person's information stored in a database and finally backpropagation neural network that can learn through a set of input and output so that the process information it perform more effectively. With these clear concepts it was chosen the analysis of main components as a method to the processing image information, because this technique is used to reduce the dimensionality of a dataset. Besides it is used to find the causes of variability of a data set and sort them by importance, I mean to a database with many variables, the objective will be reduced them to fewer number, losing the less quantity of information as possible. Developing the system based on the techniques and methods described above it was considered the C Sharp programming language for the creation of the system interface for the features provided by the management of images through its libraries that are appropriate for this type of job and Matlab for backpropagation neural network and complex mathematical calculations to be a native language in mathematical problem solving thus obtaining a robust and high performance. Ultimately this research focuses on the implementation of all the techniques described before to achieve the identification of people through the face.