拡張現実感による顔表情操作を用いたビデオ通話方式(<特集>サービス現場・日常生活に浸透するVR/AR)

抄録

Facial expression, which is one of the most important non-verbal communication media, makes communication smoother by compensating the missed message in language communication. However, some shy people are not as good at using facial expressions as they want. Such poor emotional expression by the conversation partner makes it difficult to read feelings correctly, and as the result, smooth communication is hindered. In order to solve this problem, this paper proposes a facial expression enhancement method that realizes the smooth communication with rich facial expression. To enhance facial expression, the facial shapes and textures are expressed as the parameters in parametric spaces reconstructed from the personal facial images naturally. In the parametric spaces of facial expression, the difference between two facial expressions can be handled as a multidimensional vector. By controlling the scale of the difference vector between input vector and the norm vector, the facial expression difference is enhanced without having to recognize the facial expression. Then we generate the texture, which enhanced the texture of expression by re-projecting into image space. Finally, we overlay the synthesized facial image of the conversation partner onto the face region in the video chatting sequence. We conduct on evaluations to confirm the effect of expression enhancement by our method using CG faces and a real video sequence.