Physiological Function Assessment Based On Depth Information

With the development of people's living standards,the problem for keeping physiological functions health has attracted more and more attention from the public.The decline of physiological functions has been considered as the main cause of various diseases of the human body.If proper and necessary measures are not taken,the decline in physiological function will lead to more complications to human body.Physiological function assessment is a scientific and reasonable evaluation for the capabilities of various parts of the human body,which is an important step to prevent physiological functions decline and keep human health.At present,plenty of methods has developed for evaluation to human physiological function.However,these methods needs to be performed with the full cooperation of professionals,which brings a lot of labor and time costs.In view of such problem,this paper finds that combining human activity analysis based on depth information and evaluation methods can break the inherent limitations of manual evaluation and improve the quality and efficiency of evaluation.Therefore,this paper proposes methods for the evaluation of physiological functions based on depth information.Details are as follows:1.A Kinect-based physiological function assessment method was proposed.Based on the depth information obtained from Microsoft Kinect,this paper focus on the study for the mobility of human joints from four aspects: joint mobility,motion detection,gait evaluation,and rapid upper limb assessment.2.An improved DTW algorithm is proposed.Motion detection is a necessary step to fully evaluate the physiological state of human bones and joints.In this paper,matching algorithm--DTW(Dynamic Time Warping)is used to detect upper body movements.Traditional DTW algorithms have problems such as low search efficiency and high mismatch rate.The DTW is modified from two aspects: the search path and the mapping relationship,and then this paper make a validity for MDTW by extensiveexperiments.3.A semi-automatic RULA evaluation system is proposed.RULA(Rapid Upper Limb Evaluation)is a method for assessing the risk of human motion.The evaluation process is conducted by evaluating the motion of participants from experts.Based on traditional RULA,this paper introduces Kinect for semi-automatic RULA evaluation.This system can help experts to perform effective evaluation and reduce the systematic errors introduced by manual evaluation.4.A 3D human pose estimation model based on stereo binocular camera is proposed.The Kinect depth camera developed by Microsoft can acquire 3D information of human joints in real time,but it has the disadvantages of limited number of humans,short acquisition distance,and unstable data under occlusion conditions.Therefore,in this paper,based on the 2D human pose evaluation network(Open Pose),binocular cameras are used to reconstruct joint space coordinates to obtain human joint depth information.Comparative experiments with Kinect show that this model can improve the acquisition distance of human joints and the stability of acquired joints under occlusion without affecting the accuracy of human joint acquisition.5.A physiological function assessment method based on the human posture evaluation model is proposed.In view of Kinect's shortcomings described above,this paper uses human pose estimation model to obtain human joint data containing depth information to make an evaluation for human physiological functions from two aspects: 1.TUG(Timed up and Go)assessment for human joint mobility 2.and gait assessment.Compared with Kinect V2 based physiological function evaluation method,the experimental results show that this method can be well applied to future physiological function evaluation.