Intelligent Motion Training System Based On Body Sense Recognition

The development of information technology changes all walks of life,and the growing physical health problems have also pushed the sport industry to change.Smart sport wearable devices,wisdom cloud gym and other technology products constantly impact traditional sports equipment.The latest scientific and technological achievements are combined with sports,trying to change people's exercise habits and finding new opportunities.However,it just can do some auxiliary functions such as counting the time or distance of running and swimming for the non-professional sports groups currently,and haven't provided scientific guidance for the training of body strength as the professional gym.The thesis combined Somatosensory technology with human Motion Recognition research,and put forward a set of sports training system based on somatosensory technology.According to the requirement of the system,the skeleton information is captured by the KinectV2 intelligent somatosensory device firstly,and then the skeleton data is obtained by the jitter clearing and double exponential smoothing filtering.Finally,the features between the multi-joints in the skeleton data are extracted by the space vector method.At the same time,through the process of analyzing the movement decomposition to design finite state machine.Then the finite state machine which described by the skeleton features is to be a kind of action timing reference template.A template matching algorithm based on DTW and weighted Euclidean distance is applied to realize the universal real-time motion recognition algorithm with high standard and low delay,and it can make guidance to the movements of users according to similarity calculation.This system according to "national physical training standards" and the actual situation of the campus has realized the application in three kinds of sports such as pull-ups,squat and standing long jump.At the same time,general motion recognition algorithm is used in this system and can quickly add new action in the reuse.Adhering to maintain high-quality human-computer interaction,author design and develop a real-time feedback client,and upload the movement data to cloud server through HTTP protocol.Cloud-Server support Web terminal's real-time data query and training plan making.The practical application test shows that the sports training system based on somatosensory technology proposed in this paper is stable and reliable,and adds fun and competitive to the boring sports.The research of this paper have a certain reference value for the application of somatosensory technology and the study of motion recognition algorithm.