Abstract: Non-manual component plays an important role in sign language communication. Head movement synthesis and synchronization with gesture is one of those main topics. Based on realistic training data, this paper introduces detailed explorations to reveal the relationship between gesture and head motion, and uses kernel canonical correlation analysis (KCCA) to build the head movement prediction model. Animation synthesis results and evaluation experiments imply that the proposed method can better measure the cooperation between sign gesture and head motion and enhance the naturalness of synthesized behaviors of sign language virtual human.

Key words: sign language, non-manual component, head movement, kernel canonical correlation analysis (KCCA), virtual human

摘要： 非手部手势是手语表达中不可缺少的一部分，头部运动的实现并与手势进行协同表达是其重要研究内容。对真人手语表演数据中的手势与头部动作之间的关系进行了深入研究，提取二者的动作特征，利用核典型相关分析方法（kernel canonical correlation analysis，KCCA）建立起手势与头部动作之间的预测关系模型。动画合成结果以及评价实验表明，KCCA方法能更好地刻画手势与头部动作的协调性，实现虚拟人行为动作合成的逼真性。

关键词: 手语, 非手部手势, 头部动作, 核典型相关分析, 虚拟人