3D scanned technology has found its increasing application in computer animated movie and game. Due to the over density of scanned data, simplification becomes a crucial process. Many researches have been done in mesh-based (i.e. triangulated faces) simplifications. However, few are focused on preserving the completeness of skeleton semantic. And in sharply simplification case, the losing of skeleton semantic can cause failure of skeleton based animation. Due to those problem and the time consuming of mesh reconstruction, algorithms directly based on point cloud has gained increasingly concentrations. This paper presents a simple and effective algorithm directly based on point cloud which could satisfy the succeed animation requirement when reconstructed from simplification data. It can preserve completeness of the potential skeleton semantic even in sharp decrease of points set without introducing new cracks or holes to the model when reconstructed.