Avaliação do Desempenho dos Filtros Discriminativos em um Sistema de Detecção de Pontos Fiduciais

Discriminative filtering is a pattern recognition technique which aim maximize the energy of output signal when a pattern is found. Looking improve the perfomance of filter response, was incorporated the principal component analysis in discriminative filters project. In this work we looking investigate the influence of the amount of principal components in the filter performance and by consequence the classifier applied in fiducial points detection. We showed those amount of principal components affects directly the classifier’s performance, both in the true and false positives rate.