Study on human posture recognition method based on RFID
Human posture perception is one of the major challenges in future intelligent nursing scenarios.However,vision techniques can involve issues such as occlusion,light conditions,and invasion of privacy;wearable motion tracking devices can increase the burden on human body and reduce comfort.Aiming at the above problems,a method based on radio frequency identification(RFID)for full-limb pose estimation is proposed,which can effectively overcome the problems of occlusion and wearability comfort.Firstly,based on the phase difference information,a fine-grained limb motion model is proposed,the angle of human joints is estimated through the construction of likelihood function,the forward kinematics are introduced to build multi-body joint motion model,and then combine with double antenna fusion algorithm for three-dimensional joint angle estimation.Finally,combined with human body skeleton model and body joint angle to rebuild the body posture.Experimental results using Kinect 2.0 as the standard,the measured average joint angle error is about 6°,the average joint position error is about 4 cm,indicating that the system can effectively monitor whole-body posture in real-time.
radio frequency identification(RFID)phase differencehuman body posturewearable device
NETL
NSTL
万方数据
维普
