我们采用无办法变形质量人体模板的网格从其原来的姿势对点云数据指定不同的姿势。在此方法中，我们首先建立粗模板网格和点云通过压缩的光谱嵌入技术，利用人体四肢之间的对应关系。基于这些对应关系，我们定义使用弹性能量泛函的非刚性配准的目标和应用离散的梯度流，以减少粗控制网格和点云之间的区别。然后可从之后使用均值坐标控制网格的变形变形的模板网格。(We employ a deformable, deformable human body template mesh to specify different poses from its original pose to point cloud data. In this approach, we first build coarse template meshes and point clouds through compressed spectral embedding techniques that exploit the correspondence between human limbs. Based on these correspondences, we define the use of elastic energy functionals for nonrigid registration of targets and apply discrete gradient flows to reduce the difference between coarse control grids and point clouds. Thereafter, the template mesh can be controlled by using mean coordinates to control mesh deformation.)