ResumenEl reconocimiento facial tridimensional busca subsanar las falencias que presentan los métodos basados en imágenes bidimensionales. Este tipo de reconocimiento tiene la ventaja de que las representaciones no son afectadas por cambios en la iluminación, dado que viene dada como una nube de puntos o una malla 3D donde la geometría juega un papel crucial. En este trabajo se presenta un sistema de reconocimiento de rostros, que utiliza un conjunto de descriptores de forma 3D, seleccionados a partir de un análisis de relevancia mediante coeficientes de Fisher en diferentes regiones del rostro que hacen parte de un modelo antropométrico del rostro. Se realizó un conjunto de experimentos para reconocer individuos e identificar sus expresiones y género a partir del análisis de relevancia planteado. Los resultados obtenidos muestran que la elección de características utilizando un análisis de relevancia incrementa el rendimiento del sistema de reconocimiento.Palabras clave: análisis de relevancia, aprendizaje de máquina, descriptores de forma 3D, reconocimiento de rostros 3D, segmentación 3D.
AbstractThe 3D face recognition technique aims to reduce the flaws that present the bi-dimensional based methods. This kind of recognition method has the advantage to be invariant to illumination changes because the faces are represented as a points cloud or a 3D mesh where geometry is the most remarkable feature. In this research work we present a recognition system that uses a set of 3D shape descriptors that were selected from a relevance analysis by using the Fisher coefficients in different regions of the face which are part of an anthropometric face model. A set of experiments for face, expression, and gender recognition were performed by using the relevance analysis proposed. The obtained results show that the relevance analysis increases the performance in face recognition systems.K e y w o r d s : 3 D f a c e r e c o g n i t i o n , 3 D segmentation, 3D shape descriptor, machine learning, relevance analysis.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.