Enhancing Facial Impression for Video Conference

Abstract

Most people have their preferred impression to be seen by others. Our face warping method can make this real. In this paper, we propose a new technique for automatic transformation of facial impression, for example, look more attractive or baby-face from the original. We build an impression score function, trained by scores from human raters, and the function is used to get displacement vectors for a given face. To preserve the facial identity as much as possible, we constrain the facial deformation range with facial classification. In case of real time application, such as video conference, there are frequent facial expressions variation and position changes. Face tracker is used to cope with this changing situation. Through our experiments, our proposed method can be one of the promising methods for future video conference system.