Abstract: Retargeting three-dimensional facial expressions is a hot issue in the field of computer science and technology, the animation industry and realistic social network. In this paper, we present a retargeting method of 3D facial expressions through the motion of facial feature points based on Kinect. Firstly, we use a cheap device Microsoft Kinect to capture facial motion information automatically, and build a relatively detailed 3D face model. Secondly, we use fast radial basis functions to establish the characteristic matrix, and retarget the feature point cloud from the source face model to the target face model, then drive the face motion with feature points from the source face model. Eventually retargeting facial expressions is completed in real time. Experimental results show that our method is effective and can achieve real-time requirements.