随着大量3维人体运动捕获数据库的出现，使得如何对人体运动数据进行高效分析和处理，从而有效利用运动捕获数据库成为一个新的挑战。为了高效地进行3维人体运动检索，首先通过从人体运动中提取一种基于3维空间变换特征规律的空间变换特征和运动的一些关键的时间特性来得到人体运动的3维时序特征；然后针对不同的训练需求，通过改进的数据驱动决策树的学习方法来分析关节点对运动相似的不同影响，并在检索过程中按照不同影响程度依次对关键点进行相似度计算；最终实现了一个高效的运动检索仿真系统。

With the development of motion capture techniques, more 3D motion libraries become available. In this paper, space transformation property and time property are extracted from human motion capture data, which compose 3D motion time sequence features. Given the assumption of the features of each joint is independent, a revised data-driven decision tree of anytime induction is automatically constructed to present the influence of each point during the comparison of motion similarity. Experiment results show that our approaches are effective for motion data retrieval.