Geometric feature based recognition of facial expressions using ANN

A human brain is capable of accessing the emotions by watching the facial expressions. In this paper, machine learning is proposed for six facial expressions namely anger, fear, disgust, happiness, surprise and sadness. This proposed system presents a human emotion recognition model based on facial expressions using geometrical features. The key advantage of these features is that they are independent of specific individuals. The classification system is designed by radial basis function based neural network to recognize the emotions. The system has a moderately high recognition rate of 75% with a very small feature set.