Research and Development of the Humanoid Head Portrait Robot "H&Frobot-II" with Expression and Function of Recognizing Human Expression

Abstract

A human head portrait robot system "H&Frobot-II", which has vision, expression recognition and facial expression is designed and developed in this paper. Using elastic facial skin mechanisms and driver, this system could realize that the human head portrait robot capture images of 6 basic expressions (natural, happiness, anger, sadness, surprise and disgust) from CCD camera then recognize them by expression recognition software. In the recognition system, we researched a fast algorithm to recognize human facial expression based on the facial feature value. Projection and brightness ad justness were used to extract facial expression feature points from facial image data, which were obtained by a CCD. Fuzzy neural network were used to recognition basic facial expression. And achieve the high correct recognition ratio. In this paper, we firstly proposed and researched the problem about human head portrait robot representing same-intestity expressions of human beings, and presented the concepts of expression feature points offset and the coefficient of expression feature points offset, and deduced the calculation method of same-intensity representation. Through extraction the offset of feature value, the robot can represent the facial expression. The research provide an artificially emotion model as an applicable way to express emotion.