described a
content-based cross search scheme for two kinds of three-dimensional (3D)
human motion data: time-varying mesh (TVM) and motion capture data. Chao
and colleagues [7] presented a simple and effective approach for motion
retrieval and synthesis based on posture feature indexing, and posture
features of each frame data was extracted by an index function. A 3D motion
retrieval method with motion index tree was presented by Liu and colleagues
[8]. Feature analysis is also a common used approach by many researchers.
A set of relational motion features was been defined in Demuth’s work [1].
Xiang and Zhu [9] extracted 3D temporal-spatial features of motion data and