Research On Gait Analysis System For Human Fall Risk Detection

Accidental fall,as a great threat for the health of the elderly over 65 years,causes great harm and distress to the physical and mental health of the elderly.Gait analysis,which is an effective way to detect fall risk,is increasingly becoming a research hotspot in the field of elderly health monitoring.However,there are some problems in the current research,gait analysis systems for outdoor scenes are invasive and poor applicable,and gait analysis systems for indoor scenes usually have limited detection range and limited human activity range.Considering the above problems and the increasing demand of walking aids of the elderly,a new gait analysis system is designed for human fall risk detection in this thesis,which uses a laser ranging sensor mounted on the walker to scan the movement of human legs while walking and analyze the relevant gait characteristics in a noninvasive and continuous monitoring way without restricted movement range.The main contents of this thesis include the following aspects:(1)According to the application demands of this topic,a gait analysis system is established.Data processing methods such as scanning window,leg data segmentation,and leg contour circle fitting are proposed for walker scene.Identification and tracking method for leg data segments based on Kalman filtering and difference measure is proposed to obtain the movement status of the legs.Aiming at the abnormal data frame situations such as leg occlusion and under-segmentation of leg data,a data interpolation method based on Catmull-Rom splines and a data re-segmentation method based on IEPF and geometric constraints are proposed to improve robustness of the method.(2)The zero-point constraint when walking is proposed by analyzing the variation of leg motion coordinates,and the zero-point constraint is applied to detect the gait cycle of the motion coordinate sequence to improve the traditional detection method based on peak.The acquisition method of common gait characteristics such as stride,step length,support phase,swing phase in a gait cycle is determined,(3)A more fine-grained division of the gait phase within a gait cycle is proposed,which includes 5 gait phases.The model structure is determined and the corresponding gait features are extracted according to the basic law of each gait cycle.A gait phase recognition model is established by the hidden Markov model.Aiming at the problem of parameter learning of traditional hidden Markov model,falling into the local optimal,the genetic algorithm is combined to learn the optimal parameters of the model.And the gait phase is identified by training the model and Viterbi algorithm.(4)Two application methods of gait analysis system for fall risk detection are proposed: fall risk detection based on gait feature time series and correlation rules,and fall risk detection based on gait feature information fusion.To verify the methods proposed in this thesis,different experimental schemes are designed and the experimental data is collected.The experimental results show that this method can effectively obtain the state of human legs,analyze gait characteristics such as human kinematics and periodic phase characteristics and detect the human fall risk.