Research of Tai-chi-chuan auxiliary training system based on Kinect

Abstract

In order to promote the scientific standardization of Tai-chi-chuan teaching and training, a Tai-chi-chuan auxiliary training system based on Kinect motion capture technology and extraction of the angle feature and speed feature is proposed and designed. The dynamic time planning algorithm (DTW) sampling corresponding frame is applied to replace the traditional method of sampling the key frame. The auxiliary training system consists of learning module, action acquisition module and action scoring module. Learning module with stan-dard action teaching video, to meet the requirements of students learning; action acquisition module uses Microsoft developed Kinect equipment to realize data acquisition of the spatial coordinates of the human skeleton in Tai-chi-chuan action, and through filtered noise reduction and the occlusion point data processing the skeleton data is saved. The scoring module constructs the test sequence based on the 16 angle features of eight key joint nodes and the instantaneous velocity characteristics of 20 joint nodes. By comparing with the standard sequence in real time, the original scoring rules is used to achieve the scoring function, which provides students with intuitive and reliable training recommendations, so as to improve the efficiency of learning. By increasing the number of sensors, further designing and optimizing data fusion and filtering algorithm, combining with Microsoft’s Kinect2.0 version, the system can also be applied to medical rehabilitation, physical training analysis and evaluation, virtual reality and human-computer interaction, etc.

Keywords

• image processing
• Kinect
• motion capture
• feature extraction
• DTW
• Tai-chi-chuan
• assistant training system