Abstract: A novel method is proposed to generate multimodal expressions by given emotion states.In order to generate more complex animation than Niewiadomski’s algorithm，this method constructs a practical datasets of body and hands movements for every emotional dimension.Based on the datasets and special restrains of these clips，the system searches over for several clips related to current emotion state and then blends them into multimodal expression sequences.Compared with the previous method，the expression sequences are under a high-level control thus the virtual human can act much more realistic，lifelike and fluidly emotion expressions.

Key words: emotion expression, face animation, body movement blender

摘要： 提出一种多模情感序列合成的方法。针对Niewiadomski的方法中生成动作较单一的问题，该方法细化了情感表达通道，为更多的人体部位构建所属的动作片段库，并根据运动段之间的时序与空间限制合成能够表达相关情感的动作序列。实验表明，该方法能够构建真实丰富的情感输出。

关键词: 多模情感表达, 表情动画, 人体动作合成