Research On Human Action State Discrimination Technology Based On Information Fusion

In recent years,the aging of China’s population has been increasing,and the health problems of the elderly have become a major hot spot for social research.Therefore,the health monitoring system for the elderly has been continuously optimized and improved in terms of accuracy,real-time and intelligence.To address this problem,this paper applies information fusion algorithms to the field of human body state recognition based on the traditional state recognition technology based on wearable devices.In this paper,we use acceleration sensors to collect acceleration data from different parts of the human body and fuse the acceleration data of each part node to decide the current state of the elderly.The emergence of an accurate,real-time and intelligent health monitoring system allows the family members of the elderly to monitor the current state of the elderly in real time,which can not only reduce the cost of monitoring but also effectively prevent the occurrence of accidents,while the state data collected by the sensors can also be used as a reference basis when seeking medical treatment.In a health monitoring system,the human body status data collected by a single information source sensor is inaccurate and incomplete,and does not accurately reflect the state of the target at the moment,so it is necessary to use wearable devices to collect acceleration data from different parts of the body.In this paper,purely using the acceleration sensors to constitute a wearable device,we use the wearable device to complete the acquisition of acceleration data from three parts of the human body: chest,right hand and left leg.How to fuse the information of the observed data from the sensors of different parts is a research focus of this paper.Since the acquisition time and the data returned from different sensor nodes are different,information fusion technology is needed to fuse them and accurately determine the state of the elderly.Information fusion technology can combine the work of multiple sensor acquisition nodes,synthesize data from different information sources and process them,and bring out its unique advantages in complex environments.Due to the ambiguity of human action states and the uncertainty of the boundary criteria between different states,both fuzzy theory and D-S evidence theory are effective methods to deal with the ambiguity and uncertainty of sensor observations,and can effectively eliminate the influence of erroneous data on fusion results in the fusion process.Therefore,this paper combines fuzzy theory and D-S evidence theory,and uses fuzzy theory algorithm in the feature-level fusion process to calculate the mutual support degree and fuzzy affiliation degree among each sensor observation data;in the decision-level fusion process,the support degree and affiliation degree of sensor data in different parts are transformed into the basic probability assignment values,and finally uses D-S evidence theory algorithm for information fusion,so as to fuse the sensor monitoring data according to The sensor monitoring data of different parts are used to determine whether the elderly is sitting,lying,walking or running in one of the four states at this time.The experiment proves that the human action state discrimination technology based on information fusion has good reasoning ability and can effectively make decision judgment on the state of human body for the input of acceleration data of specific parts,which has strong theoretical significance and use value,and has a certain promotion effect on the development and optimization of information fusion technology in the field of health monitoring of the elderly.