Facial expression analysis: A study on temporal models of facial action units

Abstract

There is an increased interest in developing an automatic facial expression analysis to recognize and model real human faces. In this paper, a local model-based approach for extraction of facial feature points from 2D still images and feature points tracking in image sequences are presented. It presents an algorithm that extracts the local oriented edges of intransient facial features (eyebrows, eyes, nose and mouth) to get the facial features edge map and transient features such as wrinkles densities. Finally, two experimental studies within this research paper are performed. The first one aims for measuring the accuracy of the proposed system. The second one demonstrates the temporal models of permanent and transient features.