Research On Human Motion Capture And Recognition Based On Inertial Sensor

Human motion capture systems based on inertial sensors have been widely used in movie special effects,somatosensory games,sports analysis,and medical rehabilitation due to their advantages such as convenience,economy,and practicality.However,the current human motion capture method based on inertial sensors has no external reference points and can only obtain human limb information but not the displacement information of the human body.In this paper,a human motion capture system is developed,which realizes the driving of a threedimensional human body model by an inertial sensor.Based on inertial sensor technology,combined with ultra-wideband positioning and data fusion algorithms,not only realizes fullposition human motion capture but also obtains human displacement information,making the motion capture system no longer limited to capturing the body motion of the body in place,and broadening the application range of the human motion capture system;Based on the inertial sensor technology,the ultra-wideband positioning algorithm and data fusion algorithm are combined,which not only realizes the capture of full-position human motion but also obtains the displacement information of the human body,so that the motion capture system is no longer limited to capturing the body motion of the body in place.Based on this human motion capture system,human behavior data is collected,and features are extracted.The classification accuracy rate of 16 human behaviors using classification algorithms reaches 75% to 100%,and the results show that the motion recognition effect is good.The main work of this paper is as follows:(1)Based on the inertial sensor,a human motion capture system was designed,and the corresponding software was developed.Based on human pose fusion algorithm,the real-time raw data(composed of data from the three-axis accelerometer,three-axis magnetometer,and three-axis gyroscope)in the inertial sensor is processed into quaternions,which can drive the human body model in Unity3 D.The comparison of 3D human model movements with actual human movements proves that the human motion capture system designed in this paper can accurately capture human limb movement information and achieve real-time display of 3D human models.(2)An algorithm combining inertial sensor technology and ultra-wideband positioning technology is proposed.Based on ultra-wideband positioning technology,combined with inertial sensor positioning technology.Inertial sensor-assisted positioning improves the positioning accuracy of UWB positioning technology in non-line-of-sight situations,allowing the system to obtain real-time displacement information of the human body in space during the motion capture process.The position data of the human body under non-line-of-sight conditions at rest is experimentally tested.The results show that the maximum error when using the fusion algorithm is 10 cm,and the maximum error when not using the fusion algorithm is 35 cm.By comparing human motion trajectories based on fusion and non-fusion algorithms,it is proved that the algorithm proposed in this paper can effectively improve the positioning accuracy under non-line-of-sight situations.(3)Based on the realization of human motion capture,SVM and KNN algorithms are used to recognize human behavior.Collected 16 kinds of human behavior data and extracted 9 kinds of data characteristic parameters(average value,maximum value,minimum value,standard deviation,variance,median value,extreme difference,absolute value of slope,energy).Based on the extracted feature parameters,SVM and KNN algorithms were used for human behavior recognition.The recognition rate of SVM algorithm for various human behaviors is 62.5% ~ 100%,and the recognition rate of KNN algorithm for various human behaviors is 75% ~ 100%.Among them,the two algorithms have a recognition rate of 100% for waving,crossing,picking,walking,and squatting,which can better recognize human behavior.Based on the above analysis,the SVM algorithm and KNN algorithm have a better recognition effect on human behavior,which indicates that the human motion capture system combining ultra-wideband positioning technology and inertial sensor positioning technology proposed in this paper can realize the capture and recognition of human motion,so that Human motion capture systems have become more widely used.