Caracterización y análisis comparativo de tres modelos basados en las técnicas PCA para el reconocimiento de rostros

Abstract

This project be oriented from a approach research with the purpose of studying three models of face recognition selected having as it bases specific criteria that must fulfill these models, between which we have: the model must be for face recognition, use in the recognition process the technique of Principal Components Analysis and must be use a classification system based on Artificial Neuronal Networks. The theoretical subjects related to these models are: Principal Components Analysis (PCA), Digital Processing Images (PDI), Patterns Recognitions and Artificial Neuronal Networks (ANN); these subjects studied thoroughly with the purpose of being able to implement one of the three models that are model FR_SOM, which this based on technique PCA for reduction of dimensionality and extraction of features, and uses the Artificial Neuronal Networks of Self Organized Maps. The result of the investigation we conclude that the model that uses the self organized maps by faces classification is the one that better behavior in front to presents similar models that use another type of neuronal network. This doesn’t guarantee that it is the best model of recognition of faces, but that given the case which certain specifications considered that had to fulfill the models for their study; this allowed concluding that model FR_SOM in front of presented a better result of classification the other two selected models.