中文摘要：

为能有效检索当前大量的三维人体运动捕获数据从而投入使用，针对三维人体运动捕获数据的独有特性，提出一种高效的检索方法。首先通过关键帧提取对原始运动数据降维，然后采用自定义的符合人体运动语义的特征将原始角度数据转换为特征序列。根据数据库中所有运动片段对应的特征序列构建一种基于姿态特征的索引空间。检索时通过在索引空间上顺序匹配姿态特征来查找时序一致的匹配运动片段。实验结果表明，与大多数现有方法相比，该方法由于较好的运动语义特征提取，因此具有更好的时间效率，更高的灵活性和检索精度，能够满足多种运动数据的检索需要。

英文摘要：

In order to effectively retrieve a large number of 3D human motion capture data and put it into use, this paper proposes an efficient retrieval method for the unique characteristics of 3D human motion capture data. Firstly, the original motion data dimension is reduced by key frame extraction, and then the original angle data is transformed into the feature sequence by using the customized human motion semantic feature. According to the feature sequences of all the moving segments in the database, an index space based on attitude feature is constructed. At the time of retrieval, By matching the attitude features sequentially in the index space, the matching motion fragments with the same timing are found. Experimental results show that the proposed method has better time efficiency, higher flexibility and retrieval precision than most existing methods because of better motion semantic feature extraction.