Face Emotion Recognition Using Optimality Parameters Eye and Lip in Different Geographical Area

Main Article Content

Ahmad Habibizad Navin
Mehdi Akhari ,Mirkamal Mirnia, Nima Jafari Navimipour

Abstract

The detection of emotion is becoming an important field for human – computer Interaction. Emotion recognition can be achieved by a number of methods such as facial expressions, vocal, gesture and physiology signal recognition. Facial expressions possess a number of advantages against other emotion recognition methods and suitable for a number of environmental conditions. In this paper, we describe a procedure for face emotion recognition through eye and lip optimal features. This process involves four stages: pre-processing, feature extraction, classification and comparison of emotions. Firstly a series of pre- processing tasks such as adjusting contrast, filtering, skin color segmentation and edge detection are done. One of the important tasks at this stage after pre- processing is to extract features. To extract features with high speed projection profile is used. Second particle swarm optimization (PSO) is used to optimize eye and lip ellipse characteristics. In the third stage, with using the features obtained of the optimal ellipse eye and lip, a person emotion according to experimental results and emotions represented by Ekman (sadness, angry, joy, fear, disgust and surprise without consider natural emotion) is classified. Finally, we will compare the Indian and Japanese people face emotions by PSO algorithm and also present experimental results to test. A major advantage of this method is that for each geographical area can used. To use this method in a particular geographic area, the standard parameters of the same geographic area must be set. In this case the comparison between Indian and Japanese shows that accuracy for Japanese is relatively lower than Indian.

 

Keywords: Feature extraction, Projection profile, Eye and lip ellipse, Emotions classification, Particle swarm optimization algorithm, Emotion recognition.

Downloads

Download data is not yet available.

Article Details

Section
Articles