Abstract—In this article a new real-time approach is
proposed to recognize facial expression in video database. First,
face motion direction is estimated by using motion information
and then according to the face width and direction, rectangular
surrounding box is formed. The advantage of this method
compared with previous methods is in improving classic
rectangular surrounding box model and also, using the
Steerable filters to determine face direction. As eyes and mouth
are the most important components in a face, in this method, eye
location is calculated by combining color features and
morphological functions in surrounding box region. Afterwards,
with regard to the lip contour and using image processing
techniques, we are able to determine the detection of mouth very
accurately with the help of four points. Finally, these features
are modeled by using support vector machines (SVM).
Experimental results show that this algorithm is more efficient
than the previous ones.
Index Terms—Facial expression recognition, rectangular surrounding model, support vector machines.
Sara Alipour is with the Young Researchers Club, Islamic Azad University, Qazvin Branch, Qazvin, Iran (e-mail: alipour62@ yahoo.com). Afsane Fathi is with the Islamic Azad University, Qazvin, Iran (e-mail: Afsane.email@example.com).
Cite:Sara Alipour and Afsane Fathi, "Component-Based Real Time Facial Expression Recognition in Video Streams," International Journal of Machine Learning and Computing vol.3, no. 3, pp. 309-312, 2013.