The Application Of Kinect 's Human Body Pose Recognition Method In Dance Training

Human posture recognition is the key technology of human-computer interaction. For this natural and harmony needs of human-computer interaction, in this paper, we combined the technology of human skeleton tracking based on Kinect and dancing auxiliary training, designed the dancing auxiliary training system based on Kinect. This paper proposed the skeleton joints angles representing method based on fixed axis and improved the stability on measuring these angles, then on this basis, we also improved human posture recognition method based on these angles which can identify the human postures accurately. Aiming at the condition that the skeleton joints may be occluded in human body recognition process, this paper improved the algorithm of self-occlusion information repair based on the proportion of human skeleton nodes and the motion characteristic of human body which can repair the obscured joints information accurately. We collected a large number of standard dance movements and established a standard dance movement database as the comparison template. In this way, the trainer only needs to make the performance in front of Kinect and the system can collect and recognize this movement automatically. By comparing with the standard movement data in the database, the system can evaluate trainer’s performance based on joints coordinates and angles and finally give them comparison graph and corresponding training advice.