Facial expression recognition using extended local binary patterns of 3D curvature

Abstract

This paper presents extended local binary patterns (LBP) for facial expression analysis from 3D depth map images. Recognition of facial expressions is important to understand human emotion and develop affective human computer interaction. LBP and its extensions are frequently used for texture classification and face identification and detection. In the 3D surface analysis, curvature is very important characteristics. This paper presents an extension of LBP for modeling curvature from 3D depth map images. The extended curvature LBP (CLBP) is used for facial expression recognition. Experimental results using Bosphorus facial expression database show better performance by 3D curvature and the combination of 3D curvature and 2D images than by conventional 2D or 2D + 3D approaches. © 2013 Springer Science+Business Media Dordrecht(Outside the USA).