A Probabilistic Self-Organizing Map For Facial Recognition

G. Lefebvre, C. Garcia, France Telecom R&D, 2008

This article presents a method aiming at quantifying the visual similarity between an image and a classmodel. This kind of problem is recurrent in many appli-cations such as object recognition, image classification,etc. In this paper, we propose to label a Self-OrganizingMap (SOM) to measure image similarity. To managethis goal, we feed local signatures associated to the re-gions of interest into the neural network. At the end ofthe learning step, each neural unit is tuned to a particular local signature prototype. During the labeling process, each image signature presented to the network generates an activity vote for its referent neuron. Facialrecognition is then performed by a probabilistic deci-sion rule. This scheme offers very promising results forface identification dealing with illumination variationand facial poses and expressions.