A comparative survey on supervised classifiers for face recognition

During the last decades, several different techniques have been proposed for computer recognition of human faces. A further step in the development of these biometrics is to implement them in portable devices, such as mobile phones. Due to this devices´ features and limitations it is necessary to select, among the currently available algorithms, the one with the best performance in terms of algorithm overall elapsed time and correct identification rates. The aim of this paper is to offer a complementary study to previous works, focusing on the performance of different supervised classifiers, such as the Normal Bayesian Classifier, Neural Architectures or distance-based algorithms. In addition, we analyse all the proposed algorithms´ efficiency over public face databases (ORL, FERET, NIST and the Face Recognition Data from the Essex University). Each one of these databases contains a different number of individuals and particular samples and they present variations among images from the same user (scale, pose, expression, illumination, ...). We expect to simulate many different situations which take place when dealing with face recognition on mobile phones. In order to get a complete comparison, all the proposed algorithms have been implemented and run over all the databases, using the same computer. Different parametrizations for each algorithm have also been tested. Bayesian classifiers and distance-based algorithms turn out to be the most suitable, as their parametrization is simple, the training stage is not as time consuming as others´ and classification results are satisfying.