Surface Normals with Modular Approach and Weighted Voting Scheme in 3D Facial Expression Classification

A crucial part for facial expression analysis is to capture a face deformation. In this work, we are interested by the employment of 3D facial surface normals (3DFSN) to classify six basic facial expressions and the proposed approach was employed on the Bosphorus database. We constructed a Principal Component Analysis (PCA) to capture variations in facial shape due to changes in expressions using 3DFSN as the feature vector. A modular approach is employed where a face is decomposed into six different regions and the expression classification for each module is carried out independently. We constructed a Weighted Voting Scheme (WVS) to infer the emotion underlying a collection of modules using a weight that is determined using the AdaBoost learning algorithm. Our results indicate that using 3DFSN as the feature vector of WVS yields a better performance than 3D facial points and 3D facial distance measurements in facial expression classification using both WVS and a Majority Voting Scheme (MVS). Our work is different with the existing works as they used the dataset with facial intensity information while we used dataset with no intensity. New insight in facial expression analysis is found particularly when no intensity information is provided. Surface normals does has a potential to be used as the feature vectors to classify six basic expressions. Keywords-component; Facial expression classification; 3D facial features; Principal Component Analysis; Support Vector Machines;Weighted Voting Scheme