An AAM-based Face Shape Classification Method Used for Facial Expression Recognition

Facial expression recognition is a key element in human-computer communication. However, some noises such as identity, gender and face shape may seriously have an effect on multi-person's expression recognition. In this paper, an AAM-based method is proposed to remove face shape noise for the purpose of improving facial expression recognition performance. Firstly, Active Appearance Model (AAM) is used to extract the facial feature, which includes abundant face geometry information in the form of shape parameters. Subsequently, based on the shape parameters of AAM, an SVM-based classification method is proposed to classify the main face shape--melon seed, round and square. Finally, a proposal for facial expression recognition using face shape classification is given, which can improve the recognition rate. Keywords—Active Appearance Model, facial expression recognition, face shape, SVM.

[1]  Lisa Gralewski,et al.  Statistical synthesis of facial expressions for the portrayal of emotion , 2004, GRAPHITE '04.

[2]  Simon Baker,et al.  Active Appearance Models Revisited , 2004, International Journal of Computer Vision.

[3]  Stan Sclaroff,et al.  Active blobs , 1998, Sixth International Conference on Computer Vision (IEEE Cat. No.98CH36271).

[4]  Timothy F. Cootes,et al.  Active Appearance Models , 1998, ECCV.

[5]  Ioannis Pitas,et al.  Facial Expression Recognition in Image Sequences Using Geometric Deformation Features and Support Vector Machines , 2007, IEEE Transactions on Image Processing.

[6]  T. Dalgleish,et al.  Handbook of cognition and emotion , 1999 .

[7]  Diane J. Litman,et al.  Predicting Student Emotions in Computer-Human Tutoring Dialogues , 2004, ACL.