作者: Dandan Song , Jing Dong , Qiang Zhang
DOI: 10.1007/978-3-319-26181-2_17
关键词:
摘要: The segmentation of motion capture data is essential for the synthesis data, its purpose to split long movement sequence into many different independent semantic clips, and it requires that effective accurate. This paper proposed a algorithm based on measured MDS improved oblique space distance. approach used multidimensional scaling (MDS) achieve mapping from original high-dimensional low-dimensional, then calculated distance between frames in specified windows preceding section low-dimensional space, obtained final points by similarity detection. Finally we verified feasibility through experiments, accuracy rate our method compared with traditional algorithm.