Skip to Main Content
In this paper, we propose a novel method of encoding 3D facial features for recognition. Facial structural angle (FSA) is proposed to characterize global geometric information of face. Itpsilas elements are selected by mutual information criterion. Local region map (LRM) is presented to describe salient parts around feature points on 3D face. Global (FSA) and local (LRM) descriptions are combined to play complementary roles in identification. Experimental results on GavabDB database indicate that the proposed method significantly improves recognition performance and is robust to facial expressions.