Automatic Facial Expression Recognition Using 3D Faces

Automatic facial expression recognition has gained much attention during the last decade because of its potential application in areas such as more engaging human-computer interfaces. Automatic facial expression recognition is a sub-area of face analysis research that is based heavily on methods of computer vision, machine learning, and image processing. Many efforts either to create a novel or to improve existing face expression recognition systems are thus inspired by advances in these related fields. This paper explores the automatic recognition of facial expressions using 3D range images. The paper outlines the development of an algorithm designed to distinguish between neutral and smiling faces, and summarizes its experimental verification with a database containing 30 subjects who posed for both (neutral and smiling) expressions. As a comparison with 2D farcical expression recognition, a PCA algorithm was used to exact features from 2D images and used for expression recognition. Results show that 3D facial expression recognition outperforms 2D ones.