Design And Research Of Intelligent Motion Evaluation System Based On Information Fusion Techology

With the improvement of living standards,health has become a hot issue.In order to improve the quality of sports and prevent excessive fatigue in sports,it is particularly important for people to exercise under professional guidance.Traditional sports evaluation system is mainly used in the fields of medical treatment,rehabilitation,rehabilitation and sports training.There is few evaluation systems designed for ordinary people.Therefore,this paper will use multi-level information fusion technology,combined with image tracking and recognition,wavelet transform,BP neural network and D-S evidence theory to evaluate human motion intelligently.This paper mainly studies the fatigue degree of human motion based on sensor data and video data.Firstly,motion data are acquired by Delsys sensor acquisition device through repeated experiments and motion video are taken at the same time.Secondly,an improved Gauss Mixture Background Model(GMM)algorithm and a feature point tracking algorithm are proposed to extract features from human motion video image data.Then,shi-tomasi algorithm is used to track feature points and support vector machine is used to recognize human motion posture.Thus,the fatigue degree can be judged according to the number of occurrences of different postures,and a preliminary evaluation result can be obtained.Then,the surface electromyogram data collected by the sensor are pretreated to reduce noise.Based on the wavelet transform algorithm,the surface electromyogram signals are fused at the feature level.The feature vectors are trained with BP neural network to realize the evaluation of human muscle fatigue,and a preliminary evaluation result is obtained.Finally,by studying the application of information fusion technology in decision-making level,the evaluation results of human motion comprehensive analysis are obtained by using D-S evidence theory.In addition,a web platform is built to display and manage human motion data and evaluation results,which is convenient for the storage and query of historical motion records.