A new method of motion recognition based on key frame and skeleton information is proposed. The depth sensor Kinect v2 is used to obtain the real-time information of human skeleton points

and the weighted Kmeans algorithm is used to extract the key frames in the action video. Then through the three-dimensional coordinate values of 25 skeleton points in each key frame

the joint angle and the ratio of the vector modulus are calculated. Finally

the optimized dynamic time warping（DTW） algorithm is used to calculate the similarity between the key frame sequence and the action template in the template library

so as to recognize the real time movement of the human body. The experimental results of six kinds of common movements show that this method has high recognition speed and accuracy

and has practical generalization.