中文摘要：

提出了一种自然交互方式的用户操作意图的预测算法。通过采集人体骨骼数据,建立人体手臂运动模型。采用机器视觉方式提取目标物体的特征值并建立意图模型。采用层次分析法对表达用户意图的主导因子进行权重匹配。该算法采用并行处理的方法对用户意图进行预测。通过相关实验,验证了预测算法的可靠性,该算法有助于提高人机交互的效率。

英文摘要：

A natural interactive algorithm for human intention inference was proposed. The motion model of human arm was established by collecting the data of human body skeleton. The eigenvalue of the target object was extracted by the mode of machine vision and the intention model was established. The dominant factor of user intention was matched with the weight-based pattern using the Analytic Hierarchy Process（ AHP）. The parallel processing method was used to predict the user＇s intention. The reliability of the prediction algorithm is verified by the related experiments,which can improve the efficiency of human-computer interaction.