Research On Daily Action Recognition And Fall Detection Of Elderly Based On Multi-sensor

With the deepening of the aging population of social,the elderly health care has gradually attracted attention.As the growth of the age,the elderly body begins to decline of each function,activity ability,strain capacity decline,easily in the daily life happened all sorts of dangers,and the most dangerous situation caused by the old man fall in the lower limb movement imbalances.Analysis of domestic and foreign experts and scholars on the study of human limb movement found that the vast majority of movement in life requires lower limb involvement,and therefore,the daily movement of the elderly movement model of research has a certain significance.For the study of the movement patterns of the elderly mainly through their daily activities in the body of lower limb movement patterns to implement monitoring and recognition,in order to timely find fall,avoid worsening situation.Monitoring human motion based on wireless data acquisition network is one of the hotspots in this field.On the basis of summarizing the relevant research and combining with the demand of building home health network,a portable information gathering device for limb movement of the elderly is developed.By collecting the data of different daily movements of the elderly,the data are sent to the processor through wireless transmission,and complete the various human action processing,cutting,classification preservation and identification in the processor,ultimately achieve daily actions real-time monitoring and identification for the elderly.This article designed a portable elderly limb movement information collection device.Firstly,the limb motion information acquisition is completed.The sensing device of the device includes: acquiring the three-axis acceleration information of the human body by using the acceleration sensor;the magnetic sensor acquires the inclination information of the human body and the pressure sensors collect two foot pressure data.Auxiliary hardware such as power supply,control,wireless transmission modules to realize the multiple sensor information fusion,and implementation of real-time collection of elderly movement information from various perspectives.Secondly,according to the characteristics of hardware multi-sensor information fusion in this design system,the lower-level MCU program and upper computer processing program software are designed and written in IAR ER4.21 and VC ++ 6.0 development environment respectively,so real-time communication between the sensors and the processor is realized.According to the acceleration SMV threshold algorithm and the inclination angle and pressure threshold value analysis,the daily movement data are segmented and recognized to distinguish the daily action and the fall action,and stored to the human action database,to achieve the action of data monitoring and identification,and software system in the design supplemented by human-computer interaction,alarm ambulance and other functions,to achieve fall detection.Finally,the design of daily action and typical fall test were used to test the whole system.The acceleration and inclination curves under different states of motion indicated that the system was running normally.In the end the daily movement and the typical fall were designed,and through the false positives rate of daily movement and the cognition rate of fall judgment to reflect the performance of the monitoring system.The data from several groups of experimental data shows that the system has good detection results,reliability and stability is relatively high,and achieved the goal of the project design.