Fully automatic 3D facial expression recognition using a region-based approach

Abstract

Abstract-In this paper, we propose an holistic, fully automatic approach to 3D Facial Expression Recognition (FER). A novel facial representation, namely Differential Mean Curvature Maps (DMCMs), is proposed to capture both global and local facial surface deformations which typically occur during facial expressions. These DMCMs are directly extracted from 3D depth images, by calculating the mean curvatures thanks to an integral computation. To account for facial morphology variations, they are further normalized through an aspect ratio deformation. Finally, Histograms of Oriented Gradients (HOG) are applied to regions of these normalized DMCMs and allow for the generation of facial features that can be fed to the widely used Multiclass-SVM classification algorithm. Using the protocol proposed by Gong et al.[1] on the BU-3DFE dataset, the proposed approach displays competitive performance while staying entirely automatic.

Static and dynamic 3D facial expression recognition: A comprehensive survey

“…Finally, [111] also found patches around landmarks in the face through fitting of the Statistical Facial Feature Model (SFAM), which is expressed as linear combinations of components of three different variations: shape, intensity and range value. These patches were then compared to the equivalent region from the six prototypical facial expressions through attempting to align them with ICP, and the distance between the patches after this process was used as features for classification.…”

“…In [113], the rules were discovered via ant colony and particle swarm optimisations (ACO and PCO). One of the main methods of classification that have been employed is Support Vector Machines (SVMs) [110,99,96,68,101,116,109,95,111], including multi-class SVMs [92,117]. Another technique that has been widely used is AdaBoost classification [106,99,101,109] with a selection of different weak classifiers such as linear regressors and LDA.…”

Auto-Optimized Multimodal Expression Recognition Framework Using 3D Kinect Data for ASD Therapeutic Aid

“…Lemaire et al describe a method for facial recognition that is based on localized (region-based) image analysis [3]. In this work as well as many others, the aim is to distinguish 6 primary emotions that were listed by [2]: happiness, sadness, anger, disgust, fear and surprise.…”

Sparse localized facial motion dictionary learning for facial expression recognition

“…Recently, region-or part-based approaches have been investigated in computer vision and graphics communi- ties for object and human detection [7], action recognition [8,9], video analysis [10,11], and facial motion analysis [12,13,14], and synthesis [15,16]. In the case of facial expression analysis, region-or part-based approaches have shown improvements in facial expression recognition [13,14] using manual annotation of region corresponding to local facial motion activations in different expressions.…”

“…In the case of facial expression analysis, region-or part-based approaches have shown improvements in facial expression recognition [13,14] using manual annotation of region corresponding to local facial motion activations in different expressions. In addition, region-based approaches are also supported by psycho-visual experiments that track eye-movements.…”