虚拟人情绪向量和表情向量的合成

虚拟人的面部情绪表达是一个令许多学者感兴趣的研究领域,以往的研究多偏重在人脸数据的获取和面部局部运动的生成方面,尚没有情绪合成的一般性数学模型。首先,提出了情绪向量空间的概念,人的情绪状态可以转化成情绪向量空间中的一个情绪向量,情绪向量可以通过基本情绪向量来计算。其次,建立了表情向量空间的概念,人的表情可以转化成表情向量空间中的一个表情向量,表情向量可以通过基本表情向量来计算。再次,建立从情绪向量空间到表情向量空间之间的映射,表情动画是通过情绪状态和情绪映射来实现的。最后,在微机上进行了初步的实验,利用6个基本情绪状态来计算情绪合成。

Synthesis of Emotion Vector and Expression Vector for Virtual Human

Face emotion expression of virtual human is an interesting research area for many scholars. The former research laid much stress on the acquirement of human face data and creation of the local movement on face, there is no general mathematics model for emotion synthesis. First, the concept of emotion vector was proposed and the emotion state of virtual human could be transformed to an emotion vector calculated by the basic emotion vector in emotion vector space. Second, the concept of expression vector was established and the expression of virtual human could be transformed to expression vector calculated by the basic expression vector in expression vector space. Third, the emotion mapping was set up from emotion vector space to expression vector space and emotion state and the emotion mapping implemented expression animation. Finally, a preliminary experiment was carried on microcomputer and the emotion synthesis was calculated by six basic emotions.