Research on Vision-Based Service Robot Oriented Facial Expression Recognition

Abstract

A facial expression recognition system is developed based on the robot vision to meet the facial expression recognition requirement of service robot's vision-based human-robot interaction. Firstly, the face region is localized based on skin color and Gabor texture. Secondly facial expression features are extracted according to Gabor wavelets transformation and the expression classifier is designed based on support vector machine (SVM), furthermore, seven basic facial expressions are recognized. Finally, results of experiments show that facial expressions can be real-time classified with high success rate under indoor environment by this system.