Feature Selection for Enhanced 3D Facial Expression Recognition Based on Varying Feature Point Distances

Face is the most dynamic part of the human body which comprises information about the feelings of people with facial expressions. In this paper, we propose a novel feature selection procedure applied to 3-Dimensional (3D) geometrical facial feature points selected from MPEG-4 Facial Definition Parameters (FDPs) in order to achieve robust classification performance. Distances between 3D feature point pairs are used to describe a facial expression. Support Vector Machine (SVM) is employed as the classifier. The system is tested on 3D facial expression database BU-3DFE and shows significant improvements with the proposed feature selection algorithm.